DESCRIPTION Title of Invention PRECODING METHOD, AND TRANSMITTING DEVICE Technical Field 5 [OOOl] This application is based on applications No. 2010-276457, No. 2010-293 114, NO. 201 1-035085, NO. 201 1-093543, NO. 201 1-102098, and NO. 201 1-140746 filed in Japan, the contents of which are hereby incorporated by reference. 10 [0002] The present invention relates to a precoding scheme, a precoding device, a transmission scheme, a transmission device, a reception scheme, and a reception device that in particular perform communication using a multi-antenna. Background Art 15 [0003] , f Multiple-Input Multiple-Output (MIMO) is a conventional example of a communication scheme using a multi-antenna. In multi-antenna communication, of which MIMO is representative, multiple transmission signals are each modulated, and each modulated signal is transmitted from a different antenna simultaneously in 20 order to increase the transmission speed of data. [0004] Fig. 28 shows an example of the structure of a transmission and reception device when the number of transmit antennas is two, the number of receive antennas is two, and the number of modulated signals for transmission (transmission streams) 25 is two. In the transmission device, encoded data is interleaved, the interleaved data is modulated, and fi-equency conversion and the like is performed to generate transmission signals, and the transmission signals are transmitted fi-om antennas. In this case, the scheme for simultaneously transmitting different modulated signals 1 from different transmit antennas at the same time and at the same frequency is a spatial multiplexing MIMO system. [0005] In this context, it has been suggested in Patent Literature 1 to use a 5 transmission device provided with a different interleave pattern for each transmit antenna. In other words, the transmission device in Fig. 28 would have two different interleave patterns with respective interleaves (xa, xb). As shown in Non-Patent Literature 1 and Non-Patent Literature 2, reception quality is improved in the reception device by iterative performance of a detection scheme that uses soft values 10 (the MIMO detector in Fig. 28). Models of actual propagation environments in wireless communications include non-line of sight (NLOS), of which a Rayleigh fading environment is representative, and line of sight (LOS), of which a Rician fading environment is representative. When the transmission device transmits a single modulated signal, 15 and the reception device performs maximal ratio combining on the signals received by a plurality of antennas and then demodulates and decodes the signal resulting from maximal ratio combining, excellent reception quality can be achieved in an LOS environment, in particular in an environment where the Rician factor is large, which indicates the ratio of the received power of direct waves versus the received 20 power of scattered waves. However, depending on the transmission system (for example, spatial multiplexing MIMO system), a problem occurs in that the reception quality deteriorates as the Rician factor increases (see Non-Patent Literature 3). Figs. 29A and 29B show an example of simulation results of the Bit Error Rate (BER) characteristics (vertical axis: BER, horizontal axis: signal-to-noise 25 power ratio (SNR)) for data encoded with low-density parity-check (LDPC) code and transmitted over a 2 x 2 (two transmit antennas, two receive antennas) spatial multiplexing MIMO system in a Rayleigh fading environment and in a Rician fading environment with Rician factors of K = 3, 10, and 16 dB. Fig. 29A shows the BER 2 characteristics of Max-log A Posteriori Probability (APP) without iterative detection (see Non-Patent Literature 1 and Non-Patent Literature 2), and Fig. 29B shows the BER characteristics of Max-log-APP with iterative detection (see Non-Patent Literature 1 and Non-Patent Literature 2) (number of iterations: five). As is clear 5 from Figs. 29A and 29B, regardless of whether iterative detection is performed, reception quality degrades in the spatial multiplexing MIMO system as the Rician factor increases. It is thus clear that the unique problem of "degradation of reception quality upon stabilization of the propagation environment in the spatial multiplexing MIMO system", which does not exist in a conventional single modulation signal 10 transmission system, occurs in the spatial multiplexing MIMO system. [0006] Broadcast or multicast communication is a service directed towards line-of-sight users. The radio wave propagation environment between the broadcasting station and the reception devices belonging to the users is often an 15 LOS environment. When using a spatial multiplexing MIMO system having the above problem for broadcast or multicast communication, a situation may occur in which the received electric field strength is high at the reception device, but degradation in reception quality makes it impossible to receive the service. In other words, in order to use a spatial multiplexing MIMO system in broadcast or multicast 20 communication in both an NLOS environment and an LOS environment, there is a desire for development of a MIMO system that offers a certain degree of reception quality. Non-Patent Literature 8 describes a scheme to select a codebook used in precoding (i-e. a precoding matrix, also referred to as a precoding weight matrix) 25 based on feedback information from a communication partner. Non-Patent Literature 8 does not at all disclose, however, a scheme for precoding in an environment in which feedback information cannot be acquired from the communication partner, such as in the above broadcast or multicast communication. 3 [0007] On the other hand, Non-Patent Literature 4 discloses a scheme for hopping the precoding matrix over time. This scheme can be applied even when no feedback information is available. Non-Patent Literature 4 discloses using a unitary matrix as 5 the matrix for precoding and hopping the unitary matrix at random but does not at all disclose a scheme applicable to degradation of reception quality in the above-described LOS environment. Non-Patent Literature 4 simply recites hopping between precoding matrices at random. Obviously, Non-Patent Literature 4 makes no mention whatsoever of a precoding scheme, or a structure of a precoding matrix, 10 for remedying degradation of reception quality in an LOS environment. Citation List Patent Literature [0008] Patent Literature 1 : WO 20051050885 15 Non-Patent Literature [0009] Non-Patent Literature 1 : "Achieving near-capacity on a multiple-antenna channel", IEEE Transaction on Communications, vol. 51, no. 3, pp. 389-399, Mar. 2003. 20 Non-Patent Literature 2: "Performance analysis and design optimization of LDPC-coded MIMO OFDM systems", IEEE Trans. Signal Processing, vol. 52, no. 2, pp. 348-361, Feb. 2004. Non-Patent Literature 3: "BER performance evaluation in 2 x 2 MIMO spatial multiplexing systems under Rician fading channels", IEICE Trans. 25 Fundamentals, vol. E91-A, no. 10, pp. 2798-2807, Oct. 2008. Non-Patent Literature 4: "Turbo space-time codes with time varying linear transformations", IEEE Trans. Wireless communications, vol. 6, no. 2, pp. 486493, Feb. 2007. 4 Non-Patent Literature 5: "Likelihood function for QR-MLD suitable for soft-decision turbo decoding and its performance", IEICE Trans. Commun., vol. E88-B, no. 1, pp. 47-57, Jan. 2004. Non-Patent Literature 6: "A tutorial on 'parallel concatenated (Turbo) coding', 'Turbo (iterative) decoding' and related topics", The Institute of Electronics, Information, and Communication Engineers, Technical Report IT 98-5 1. Non-Patent Literature 7: "Advanced signal processing for PLCs: Wavelet-OFDM", Proc. of IEEE International symposium on ISPLC 2008, pp.187-192,2008. Non-Patent Literature 8: D. J. Love, and R. W. Heath, Jr., "Limited feedback unitary precoding for spatial multiplexing systems", IEEE Trans. Inf. Theory, vol. 51, no. 8, pp. 2967-2976, Aug. 2005. Non-Patent Literature 9: DVB Document A122, Framing structure, channel coding and modulation for a second generation digital terrestrial television broadcasting system, (DVB-T2), Jun. 2008. Non-Patent Literature 10: L. Vangelista, N. Benvenuto, and S. Tomasin, "Key technologies for next-generation terrestrial digital television standard DVB-T2", IEEE Commun. Magazine, vol. 47, no. 10, pp. 146-153, Oct. 2009. Non-Patent Literature 11: T. Ohgane, T. Nishimura, and Y. Ogawa, "Application of space division multiplexing and those performance in a MIMO channel", IEICE Trans. Commun., vol. 88-B, no. 5, pp. 1843-1 85 1, May 2005. Non-Patent Literature 12: R. G. Gallager, "Low-density parity-check codes", IRE Trans. Inform. Theory, IT-8, pp. 21-28, 1962. Non-Patent Literature 13: D. J. C. Mackay, "Good error-correcting codes based on very sparse matrices", IEEE Trans. Inform. Theory, vol. 45, no. 2, pp. 399-43 1, March 1999. Non-Patent Literature 14: ETSI EN 302 307, "Second generation framing structure, channel coding and modulation systems for broadcasting, interactive 5 services, news gathering and other broadband satellite applications", v. 1.1.2, June 2006. Non-Patent Literature 15: Y.-L. Ueng, and C.-C. Cheng, "A fast-convergence decoding method and memory-efficient VLSI decoder architecture 5 for irregular LDPC codes in the IEEE 802.16e standards", IEEE VTC-2007 Fall, pp. 1255-1259. Summary of Invention Technical Problem [OO lo] 10 It is an object of the present invention to provide a MIMO system that improves reception quality in an LOS environment. Solution to Problem [OO 1 11 To solve the above problem, the present invention provides a precoding 15 method for generating, from a plurality of signals which are based on a selected modulation scheme and represented by in-phase components and quadrature components, a plurality of precoded signals that are transmitted in the same frequency bandwidth at the same time and transmitting the generated precoded signals, the precoding method comprising: selecting one precoding weight matrix 20 from among a plurality of precoding weight matrices by regularly hopping between the matrices; and generating the plurality of precoded signals by multiplying the selected precoding weight matrix by the plurality of signals which are based on the selected modulation scheme, the plurality of precoding weight matrices being nine matrices expressed, using a positive real number a, as Equations 339 through 347 25 (details are described below). [OO 121 According to each aspect of the above invention, precoded signals, which are generated by precoding signals by using one precoding weight matrix selected 6 from among a plurality of precoding weight matrices by regularly hopping between the matrices, are transmitted and received. Thus the precoding weight matrix used in the precoding is any of a plurality of precoding weight matrices that have been predetermined. This makes it possible to improve the reception quality in an LOS 5 environment based on the design of the plurality of precoding weight matrices. Advantageous Effects of Invention [0013] With the above structure, the present invention provides a precoding method, a precoding device, a transmission method, a reception method, a transmission 10 device, and a reception device that remedy degradation of reception quality in an LOS environment, thereby providing high-quality service to LOS users during broadcast or multicast communication. Brief Description of Drawings [00 1 41 15 Fig. 1 is an example of the structure of a transmission device and a reception device in a spatial multiplexing MIMO system. Fig. 2 is an example of a frame structure. Fig. 3 is an example of the structure of a transmission device when adopting a scheme of hopping between precoding weights. 20 Fig. 4 is an example of the structure of a transmission device when adopting a scheme of hopping between precoding weights. Fig. 5 is an example of a frame structure. Fig. 6 is an example of a scheme of hopping between precoding weights. Fig. 7 is an example of the structure of a reception device. 25 Fig. 8 is an example of the structure of a signal processing unit in a reception device. Fig. 9 is an example of the structure of a signal processing unit in a reception device. 7 Fig. 10 shows a decoding processing scheme. Fig. 1 1 is an example of reception conditions. Figs. 12A and 12B are examples of BER characteristics. Fig. 13 is an example of the structure of a transmission device when 5 adopting a scheme of hopping between precoding weights. Fig. 14 is an example of the structure of a transmission device when adopting a scheme of hopping between precoding weights. Figs. 15A and 15B are examples of a frame structure. Figs. 16A and 16B are examples of a frame structure. Figs. 17A and 17B are examples of a frame structure. Figs. 18A and 18B are examples of a frame structure. Figs. 19A and 19B are examples of a frame structure. Fig. 20 shows positions of poor reception quality points. Fig. 2 1 shows positions of poor reception quality points. Fig. 22 is an example of a frame structure. Fig. 23 is an example of a frame structure. Figs. 24A and 24B are examples of mapping schemes. Figs. 25A and 25B are examples of mapping schemes. Fig. 26 is an example of the structure of a weighting unit. Fig. 27 is an example of a scheme for reordering symbols. Fig. 28 is an example of the structure of a transmission device and a reception device in a spatial multiplexing MIMO system. Figs. 29A and 29B are examples of BER characteristics. Fig. 30 is an example of a 2 x 2 MIMO spatial multiplexing MIMO system. Figs. 3 1A and 3 1B show positions of poor reception points. Fig. 32 shows positions of poor reception points. Figs. 33A and 33B show positions of poor reception points. Fig. 34 shows positions of poor reception points. 8 Figs. 35A and 35B show positions of poor reception points. Fig. 36 shows an example of minimum distance characteristics of poor reception points in an imaginary plane. Fig. 37 shows an example of minimum distance characteristics of poor 5 reception points in an imaginary plane. Figs. 38A and 38B show positions of poor reception points. Figs. 39A and 39B show positions of poor reception points. Fig. 40 is an example of the structure of a transmission device in Embodiment 7. Fig. 41 is an example of the frame structure of a modulated signal transmitted by the transmission device. Figs. 42A and 42B show positions of poor reception points. Figs. 43A and 43B show positions of poor reception points. Figs. 44A and 44B show positions of poor reception points. Figs. 45A and 45B show positions of poor reception points. Figs. 46A and 46B show positions of poor reception points. Figs. 47A and 47B are examples of a fiame structure in the time and fiequency domains. Figs. 48A and 48B are examples of a fiame structure in the time and 20 fiequency domains. Fig. 49 shows a signal processing scheme. Fig. 50 shows the structure of modulated signals when using space-time block coding. Fig. 51 is a detailed example of a fiame structure in the time and frequency 25 domains. Fig. 52 is an example of the structure of a transmission device. Fig. 53 is an example of a structure of the modulated signal generating units #I-#M in Fig. 52. 9 Fig. 54 shows the structure of the OFDM related processors (5207-1 and 5207-2) in Fig. 52. Figs. 55A and 55B are detailed examples of a frame structure in the time and frequency domains. 5 Fig. 56 is an example of the structure of a reception device. Fig. 57 shows the structure of the OFDM related processors (5600-X and 5600-Y) in Fig. 56. Figs. 58A and 5 8a~re detailed examples of a frame structure in the time and frequency domains. Fig. 59 is an example of a broadcasting system. Figs. 60A and 60B show positions of poor reception points. Fig. 61 is an example of the frame structure. Fig. 62 is an example of a frame structure in the time and frequency domain. Fig. 63 is an example of a structure of a transmission device. Fig. 64 is an example of a frame structure in the frequency and time domain. Fig. 65 is an example of the h e structure. Fig. 66 is an example of symbol arrangement scheme. 20 Fig. 67 is an example of symbol arrangement scheme. Fig. 68 is an example of symbol arrangement scheme. Fig. 69 is an example of the frame structure. Fig. 70 shows a frame structure in the time and frequency domain. Fig. 71 is an example of a h e structure in the time and frequency 25 domain. Fig. 72 is an example of a structure of a transmission device. Fig. 73 is an example of a structure of a reception device. Fig. 74 is an example of a structure of a reception device. 10 Fig. 75 is an example of a structure of a reception device. Figs. 76A and 76B show examples of a h e structure in a frequency-time domain. Figs. 77A and 77B show examples of a h e structure in a frequency-time 5 domain. Figs. 78A and 78B show a result of allocating precoding matrices. Figs. 79A and 79B show a result of allocating precoding matrices. Figs. 80A and 80B show a result of allocating precoding matrices. Fig. 81 is an example of the structure of a signal processing unit. Fig. 82 is an example of the structure of a signal processing unit. Fig. 83 is an example of the structure of the transmission device. Fig. 84 shows the overall structure of a digital broadcasting system. Fig. 85 is a block diagram showing an example of the structure of a reception device. Fig. 86 shows the structure of multiplexed data. Fig. 87 schematically shows how each stream is multiplexed in the multiplexed data. Fig. 88 shows in more detail how a video stream is stored in a sequence of PES packets. 20 Fig. 89 shows the structure of a TS packet and a source packet in multiplexed data. Fig. 90 shows the data structure of a PMT. Fig. 91 shows the internal structure of multiplexed data information. Fig. 92 shows the internal structure of stream attribute information. Fig. 93 is a structural diagram of a video display and an audio output device. Fig. 94 is an example of signal point layout for 16QAM. Fig. 95 is an example of signal point layout for QPSK. 11 Fig. 96 shows a baseband signal hopping unit. Fig. 97 shows the number of symbols and the number of slots. Fig. 98 shows the number of symbols and the number of slots. Figs. 99A and 99B each show a structure of a frame structure. Fig. 100 shows the number of slots. Fig. 101 shows the number of shots. Fig. 102 shows a PLP in the time and frequency domain. Fig. 103 shows a structure of the PLP. Fig. 104 shows a PLP in the time and frequency domain. Fig. 105 schematically shows absolute values of a log-likelihood ratio obtained by the reception device. Fig. 106 schematically shows absolute values of a log-likelihood ratio obtained by the reception device. Fig. 107 is an example of a structure of a signal processing unit pertaining 15 to a weighting combination unit. Fig. 108 is an example of a structure of the signal processing unit pertaining to the weighting combination unit. Fig. 109 is an example of signal point layout in the I-Q plane for 64QAM. Fig. 110 shows a chart pertaining to the precoding matrices. Fig. 11 1 shows a chart pertaining to the precoding matrices. Fig. 112 is an example of a structure of the signal processing unit pertaining to the weighting combination unit. Fig. 113 is an example of a structure of the signal processing unit pertaining to the weighting combination unit. Fig. 114 shows a chart pertaining to the precoding matrices. Fig. 1 15 shows a chart pertaining to the precoding matrices. Fig. 116 is an example of a structure of the signal processing unit pertaining to the weighting combination unit. 12 Fig. 117 is an example of signal point layout. Fig. 1 18 shows a relationship of positions of signal points. Fig. 119 is an example of signal point layout. Fig. 120 is an example of a structure of a signal generating unit. 5 Fig. 121 shows in-phase components and quadrature components of baseband signals. Fig. 122 is an example of a structure of the signal generating unit. Fig. 123 is an example of a structure of the signal generating unit. Fig. 124 shows in-phase components and quadrature components of 10 baseband signals. Fig. 125 is an example of a structure of the signal generating unit. Fig. 126 is an example of a structure of the signal generating unit. Description of Embodiments [00 1 51 15 The following describes embodiments of the present invention with reference to the drawings. (Embodiment 1) The following describes the transmission scheme, transmission device, reception scheme, and reception device of the present embodiment. 20 [0016] Prior to describing the present embodiment, an overview is provided of a transmission scheme and decoding scheme in a conventional spatial multiplexing MIMO system. Fig. 1 shows the structure of an Nt x N, spatial multiplexing MIMO system. 25 An information vector z is encoded and interleaved. As output of the interleaving, an encoded bit vector u = (ul, . . ., uNt) is acquired. Note that u, = (uIl, . . ., u& (where M is the number of transmission bits per symbol). Letting the transmission vector s = (sl, . . .. , sNtlTa nd the transmission signal fiom transmit antenna #1 be represented as 13 s, = map(ui), the normalized transmission energy is represented as ~{ls~l=* )E s/Nt (E, being the total energy per channel). Furthermore, letting the received vector be y = (yl, .. . , yNr)Tt,h e received vector is represented as in Equation 1. [OO 1 71 5 Math 1 Equation 1 [OO 1 81 10 In this Equation, HNmr is the channel matrix, n = (nl, . . ., nNdT is the noise vector, and n, is the i.i.d. complex Gaussian random noise with an average value 0 and variance 2. From the relationship between transmission symbols and reception symbols that is induced at the reception device, the probability for the received vector may be provided as a multi-dimensional Gaussian distribution, as in Equation 15 2. [00 1 91 Math 2 Equation 2 [0020] Here, a reception device that performs iterative decoding composed of an outer soft-idsoft-out decoder and a MIMO detector, as in Fig. 1, is considered. The 14 vector of a log-likelihood ratio (L-value) in Fig. 1 is represented as in Equations 3-5. [002 11 Math 3 5 Equation 3 [0022] Math 4 10 Equation 4 [0023] Math 5 15 Equation 5 [0024] 20 The following describes iterative detection of MIMO signals in the N, x N, spatial multiplexing MIMO system. The log-likelihood ratio of u, is defined as in Equation 6. 15 [0025] Math 6 Equation 6 5 [0026] From Bayes' theorem, Equation 6 can be expressed as Equation 7. [0027] Math 7 10 Equation 7 [0028] Let = {ulu, = *I)- When approximating I d a j - ma. In aj, an 15 approximation of Equation 7 can be sought as Equation 8. Note that the above symbol "-" indicates approximation. [0029] Math 8 Equation 8 5 P(ulu,) and In P(ulu,) in Equation 8 are represented as follows. [003 11 Math 9 Equation 9 = n " rv)*(-n) exp[~(yq)] + exp [ - "(:.I) [0032] Math 10 Equation 10 Math 11 Equation 11 5 [0034] Incidentally, the logarithmic probability of the equation defined in Equation 2 is represented in Equation 12. [0035] Math 12 10 Equation 12 [0036] Accordingly, fiom Equations 7 and 13, in MAP or A Posteriori Probability 15 (APP), the a posteriori L-value is represented as follows. [0037] Math 13 Equation 13 [003 81 Hereinafter, this is referred to as iterative APP decoding. From Equations 8 5 and 12, in the log-likelihood ratio utilizing Max-Log approximation (Max-Log APP), the a posteriori L-value is represented as follows. [0039] Math 14 Equation 14 10 L(urnI Yn) % Umman,x+l { ~ ( uy, ~ ( u ) )-) U mn,-1 { ~ ( uY,, ~ (u))} [0040] Math 15 Equation 15 [004 11 Hereinafter, this is referred to as iterative Max-log APP decoding. The extrinsic information required in an iterative decoding system can be sought by I I 20 subtracting prior inputs from Equations 13 and 14. Fig. 28 shows the basic structure of the system that is related to the subsequent description. This system is a 2 x 2 spatial multiplexing MIMO system. There is an outer encoder for each of streams A and B. The two outer encoders are identical LDPC encoders. (Here, a structure using LDPC encoders as the outer 5 encoders is described as an example, but the error correction coding used by the outer encoder is not limited to LDPC coding. The present invention may similarly be embodied using other error correction coding such as turbo coding, convolutional coding, LDPC convolutional coding, and the like. Furthermore, each outer encoder is described as having a transmit antenna, but the outer encoders are not limited to 10 this structure. A plurality of transmit antennas may be used, and the number of outer encoders may be one. Also, a greater number of outer encoders may be used than the number of transmit antennas.) The streams A and B respectively have interleavers (n, q,). Here, the modulation scheme is 2 h -(w~ith h~ bit~s tra nsmitted in one symbol). 15 The reception device performs iterative detection on the above MIMO signals (iterative APP (or iterative Max-log APP) decoding). Decoding of LDPC codes is performed by, for example, sum-product decoding. Fig. 2 shows a frame structure and lists the order of symbols after interleaving. In this case, (i, j,), (ib, jb) are represented by the following Equations. 20 [0042] Math 16 Equation 16 25 [0043] Math 17 Equation 17 [0044] In this case, ia, ib indicate the order of symbols after interleaving, ja, jb 5 indicate the bit positions (ja, jb = 1, . . ., h) in the modulation scheme, xa, xb indicate the interleavers for the streams A and B, and RaiqJ, Rb,b, Jb indicate the order of data in streams A and B before interleaving. Note that Fig. 2 shows the frame structure for ia = ib. 10 The following is a detailed description of the algorithms for sum-product decoding used in decoding of LDPC codes and for iterative detection of MIMO signals in the reception device. [0045] Sum-Product Decoding 15 Let a two-dimensional M x N matrix H = {H,) be the check matrix for LDPC codes that are targeted for decoding. Subsets A(m), B(n) of the set [l, N] = (1, 2, . . ., N) are defined by the following Equations. [0046] Math 18 20 Equation 18 [0047] Math 19 25 Equation 19 [0048] In these Equations, A(m) represents the set of column indices of 1's in the 5 mh column of the check matrix H, and B(n) represents the set of row indices of 1's in the nh row of the check matrix H. The algorithm for sum-product decoding is as follows. Step As1 (initialization): let a priori value log-likelihood ratio P, = 0 for all combinations (m, n) satisfying H, = 1. Assume that the loop variable (the number 10 of iterations) I, = 1 and the maximum number of loops is set to l,, ,,. Step A.2 (row processing): the extrinsic value log-likelihood ratio a, is updated for all combinations (m, n) satisfying H, = 1 in the order of m = 1, 2, . . ., M, using the following updating Equations. [0049] 15 Math 20 Equation 20 [0050] 20 Math 21 Equation 21 [005 11 Math 22 Equation 22 exp(x) + 1 f (x) = In exp(x) - 1 5 [0052] In these Equations, f represents a Gallager function. Furthermore, the scheme of seeking I,, is described in detail later. Step A-3 (column processing): the extrinsic value log-likelihood ratio P, is updated 10 for all combinations (m, n) satisQing H,, = 1 in the order of n = 1, 2, . . ., N, using the following updating Equation. [0053] Math 23 Equation 23 [0054] Step A-4 (calculating a log-likelihood ratio): the log-likelihood ratio I, is sought for n E [I, N] by the following Equation. 20 [0055] Math 24 Equation 24 [0056] Step A.5 (count of the number of iterations): if I,, ,I, <, I,,,,,,,, then I, is 5 incremented, and processing returns to step A-2. If 1,- ,=, I, the sum-product decoding in this round is finished. The operations in one sum-product decoding have been described. Subsequently, iterative MIMO signal detection is performed. In the variables m, n, a, p, h,, and L, used in the above description of the operations of sum-product 10 decoding, the variables in stream A are m, n, a"-, Pa-, h,, and L, and the variables in s~~eaBm ar e mb,nb, clbmbnb,p bmbnbh ,b and Lnb. The following describes the scheme of seeking h, in iterative MIMO signal detection in detail. 15 [0057] The following Equation holds from Equation 1. [0058] Math 25 Equation 25 [0059] The following Equations are defined fi-om the fiame structures of Fig. 2 and from Equations 16 and 17. 24 [0060] Math 26 Equation 26 5 [006 11 Math 27 Equation 27 b nb = n i b , jb 10 [0062] In this case, n,nb E [I, N]. Hereinafter, L, Lb, and Lnb, where the number of iterations of iterative MIMO signal detection is k, are represented as hk , Lk, na, h, n b and Lk, nb- 15 [0063] Step B-1 (initial detection; k = 0): b,, and k,,b are sought as follows in the case of initial detection. In iterative APP decoding: [0064] 20 Math 28 Equation 28 [0065] In iterative Max-log APP decoding: 5 [0066] Math 29 Equation 29 10 [0067] Math 30 Equation 30 15 [0068] Here, let X = a, b. Then, assume that the number of iterations of iterative MIMO signal detection is = 0 and the maximum number of iterations is set to Step B-2 (iterative detection; the number of iterations k): h,, and kk nb, where the number of iterations is k, are represented as in Equations 31-34, fiom Equations 1 1, 13-1 5, 16, and 17. Let (X, Y) = (a, b)(b, a). In iterative APP decoding: 5 [0070] Math 3 1 Equation 3 1 10 [0071] Math 32 Equation 32 15 100721 In iterative Max-log APP decoding: [0073] Math 33 Equation 33 [0074] Math 34 Equation 34 [0075] Step B-3 (counting the number of iterations and estimating a codeword): increment lmimiof lmim The INNER MIMO detector 803 receives, as inputs, the baseband signal 801X, the channel estimation signal group 802X, the baseband signal 801Y, and the channel estimation signal group 802Y. Here, the modulation scheme for the 20 modulated signal (stream) sl and the modulated signal (stream) s2 is described as 16QAM. [0121] The INNER MlMO detector 803 first calculates H(t)W(t) fiom the channel estimation signal group 802X and the channel estimation signal group 802Y to seek 25 candidate signal points corresponding to the baseband signal 801X. Fig. 11 shows such calculation. In Fig. 11, each black dot (e) is a candidate signal point in the I-Q plane. Since the modulation scheme is 16QAM, there are 256 candidate signal points. (Since Fig. 11 is only for illustration, not all 256 candidate signal points are 40 shown.) Here, letting the four bits transferred by modulated signal sl be bO, bl, b2, and b3, and the four bits transferred by modulated signal s2 be b4, b5, b6, and b7, candidate signal points corresponding to (bO, bl, b2, b3, b4, b5, b6, b7) in Fig. 11 exist. The squared Euclidian distance is sought between a received signal point 1 101 5 (corresponding to the baseband signal 801X) and each candidate signal point. Each squared Euclidian distance is divided by the noise variance c?. Accordingly, Ex(bO, bl, b2, b3, b4, b5, b6, b7), i.e. the value of the squared Euclidian distance between a candidate signal point corresponding to (bO, bl, b2, b3, b4, b5, b6, b7) and a received signal point, divided by the noise variance, is sought. Note that the 10 baseband signals and the modulated signals sl and s2 are each complex signals. [O 1221 Similarly, H(t)W(t) is calculated fkom the channel estimation signal group 802X and the channel estimation signal group 802Y, candidate signal points corresponding to the baseband signal 801Y are sought, the squared Euclidian 15 distance for the received signal point (corresponding to the baseband signal 801Y) is sought, and the squared Euclidian distance is divided by the noise variance 02. Accordingly, EY(bO, bl, b2, b3, b4, b5, b6, b7), i.e. the value of the squared Euclidian distance between a candidate signal point corresponding to (bO, bl, b2, b3, b4, b5, b6, b7) and a received signal point, divided by the noise variance, is sought. 20 [0123] Then Ex(bO, bl, b2, b3, b4, b5, b6, b7) + Ey(bO, bl, b2, b3, b4, b5, b6, b7) = E(b0, bl, b2, b3, b4, b5, b6, b7) is sought. [O 1241 The INNER MlMO detector 803 outputs E(b0, bl, b2, b3, b4, b5, b6, b7) as 25 a signal 804. [0 1251 A log-likelihood calculating unit 805A receives the signal 804 as input, calculates the log likelihood for bits bO, bl, b2, and b3, and outputs a log-likelihood 41 signal 806A. Note that during calculation of the log likelihood, the log likelihood for "1" and the log likelihood for "0" are calculated. The calculation scheme is as shown in Equations 28, 29, and 30. Details can be found in Non-Patent Literature 2 and Non-Patent Literature 3. 5 [0126] Similarly, a log-likelihood calculating unit 805B receives the signal 804 as input, calculates the log likelihood for bits b4, b5, b6, and b7, and outputs a log-likelihood signal 806B. [0 1271 10 A deinterleaver (807A) receives the log-likelihood signal 806A as an input, performs deinterleaving corresponding to the interleaver (the interleaver (304A) in Fig. 3), and outputs a deinterleaved log-likelihood signal 808A. [0 1281 Similarly, a deinterleaver (807B) receives the log-likelihood signal 806B as 15 an input, performs deinterleaving corresponding to the interleaver (the interleaver (304B) in Fig. 3), and outputs a deinterleaved log-likelihood signal 808B. [0 1291 A log-likelihood ratio calculating unit 809A receives the interleaved log-likelihood signal 808A as an input, calculates the log-likelihood ratio (LLR) of 20 the bits encoded by the encoder 302A in Fig. 3, and outputs a log-likelihood ratio signal 810A. [0130] Similarly, a log-likelihood ratio calculating unit 809B receives the interleaved log-likelihood signal 808B as an input, calculates the log-likelihood ratio 25 (LLR) of the bits encoded by the encoder 302B in Fig. 3, and outputs a log-likelihood ratio signal 8 10B. [0131] A soft-idsoft-out decoder 81 1A receives the log-likelihood ratio signal 810A as an input, performs decoding, and outputs a decoded log-likelihood ratio 812A. [0132] 5 Similarly, a soft-idsoft-out decoder 81 1B receives the log-likelihood ratio signal 810B as an input, performs decoding, and outputs a decoded log-likelihood ratio 812B. [0133] 10 An interleaver (813A) receives the log-likelihood ratio 812A decoded by the soft-inlsoft-out decoder in the (k - I ) i~ter ation as an input, performs interleaving, and outputs an interleaved log-likelihood ratio 814A. The interleaving pattern in the interleaver (813A) is similar to the interleaving pattern in the interleaver (304A) in Fig. 3. 15 [0134] An interleaver (8 1 3B) receives the log-likelihood ratio 8 12B decoded by the soft-idsoft-out decoder in the (k - 1 )i~ter ation as an input, performs interleaving, and outputs an interleaved log-likelihood ratio 814B. The interleaving pattern in the interleaver (813B) is similar to the interleaving pattern in the interleaver (304B) in 20 Fig. 3. [0135] The INNER MIMO detector 803 receives, as inputs, the baseband signal 81 6X, the transformed channel estimation signal group 81 7X, the baseband signal 816Y, the transformed channel estimation signal group 817Y, the interleaved 25 log-likelihood ratio 814A, and the interleaved log-likelihood ratio 814B. The reason for using the baseband signal 8 16X, the transformed channel estimation signal group 817X, the baseband signal 816Y, and the transformed channel estimation signal group 817Y instead of the baseband signal 801X, the channel estimation signal 43 group 802X, the baseband signal 801Y, and the channel estimation signal group 802Y is because a delay occurs due to iterative decoding. [0136] The difference between operations by the INNER MIMO detector 803 for 5 iterative decoding and for initial detection is the use of the interleaved log-likelihood ratio 814A and the interleaved log-likelihood ratio 814B during signal processing. The INNER MIMO detector 803 first seeks E(b0, bl, b2, b3, b4, b5, b6, b7), as during initial detection. Additionally, coefficients corresponding to Equations 11 and 32 are sought fiom the interleaved log-likelihood ratio 814A and the interleaved 10 log-likelihood ratio 914B. The value E(b0, bl, b2, b3, b4, b5, b6, b7) is adjusted using the sought coefficients, and the resulting value E'(b0, bl, b2, b3, b4, b5, b6, b7) is output as the signal 804. [0137] The log-likelihood calculating unit 805A receives the signal 804 as input, 15 calculates the log likelihood for bits bO, bl, b2, and b3, and outputs the log-likelihood signal 806A. Note that during calculation of the log likelihood, the log likelihood for "1" and the log likelihood for "0" are calculated. The calculation scheme is as shown in Equations 31, 32, 33, 34, and 35. Details can be found in Non-Patent Literature 2 and Non-Patent Literature 3. 20 [0138] Similarly, the log-likelihood calculating unit 805B receives the signal 804 as input, calculates the log likelihood for bits b4, b5, b6, and b7, and outputs the log-likelihood signal 806B. Operations by the deinterleaver onwards are similar to initial detection. 25 [0139] Note that while Fig. 8 shows the structure of the signal processing unit when performing iterative detection, iterative detection is not always essential for obtaining excellent reception quality, and a structure not including the interleavers 44 813A and 813B, which are necessary only for iterative detection, is possible. In such a case, the INNER MIMO detector 803 does not perform iterative detection. The main part of the present embodiment is calculation of H(t)W(t). Note that as shown in Non-Patent Literature 5 and the like, QR decomposition may be 5 used to perform initial detection and iterative detection. [0 1401 Furthermore, as shown in Non-Patent Literature 11, based on H(t)W(t), linear operation of the Minimum Mean Squared Error (MMSE) and Zero Forcing (ZF) may be performed in order to perform initial detection. 10 [0141] Fig. 9 is the structure of a different signal processing unit than Fig. 8 and is for the modulated signal transmitted by the transmission device in Fig. 4. The difference with Fig. 8 is the number of soft-idsoft-out decoders. A soft-idsoft-out decoder 901 receives, as inputs, the log-likelihood ratio signals 810A and 810B, 15 performs decoding, and outputs a decoded log-likelihood ratio 902. A distribution unit 903 receives the decoded log-likelihood ratio 902 as an input and distributes the log-likelihood ratio 902. Other operations are similar to Fig. 8. [0 1421 Figs. 12A and 12B show BER characteristics for a transmission scheme 20 using the precoding weights of the present embodiment under similar conditions to Figs. 29A and 29B. Fig. 12A shows the BER characteristics of Max-log A Posteriori Probability (APP) without iterative detection (see Non-Patent Literature 1 and Non-Patent Literature 2), and Fig. 12B shows the BER characteristics of Max-log-APP with iterative detection (see Non-Patent Literature 1 and Non-Patent 25 Literature 2) (number of iterations: five). Comparing Figs. 12A, 12B, 29A, and 29B shows how if the transmission scheme of the present embodiment is used, the BER characteristics when the Rician factor is large greatly improve over the BER characteristics when using spatial multiplexing MIMO system, thereby confirming the usefulness of the scheme in the present embodiment. [0143] As described above, when a transmission device transmits a plurality of 5 modulated signals from a plurality of antennas in a MIMO system, the advantageous effect of improved transmission quality, as compared to conventional spatial multiplexing MIMO system, is achieved in an LOS environment in which direct waves dominate by hopping between preceding weights regularly over time, as in the present embodiment. 10 [0144] In the present embodiment, and in particular with regards to the structure of the reception device, operations have been described for a limited number of antennas, but the present invention may be embodied in the same way even if the number of antennas increases. In other words, the number of antennas in the 15 reception device does not affect the operations or advantageous effects of the present embodiment. Furthermore, in the present embodiment, the example of LDPC coding has particularly been explained, but the present invention is not limited to LDPC coding. Furthermore, with regards to the decoding scheme, the soft-inlsoft-out decoders are not limited to the example of sum-product decoding. Another 20 soft-inlsoft-out decoding scheme may be used, such as a BCJR algorithm, a SOVA algorithm, a Max-log-MAP algorithm, and the like. Details are provided in Non-Patent Literature 6. [0 1 451 Additionally, in the present embodiment, the example of a single carrier 25 scheme has been described, but the present invention is not limited in this way and may be similarly embodied for multi-carrier transmission. Accordingly, when using a scheme such as spread spectrum communication, Orthogonal Frequency-Division Multiplexing (OFDM), Single Carrier Frequency Division Multiple Access 46 (SC-FDMA), Single Carrier Orthogonal Frequency-Division Multiplexing (SC-OFDM), or wavelet OFDM as described in Non-Patent Literature 7 and the like, for example, the present invention may be similarly embodied. Furthermore, in the present embodiment, symbols other than data symbols, such as pilot symbols 5 (preamble, unique word, and the like), symbols for transmission of control information, and the like, may be arranged in the frame in any way. [O 1461 The following describes an example of using OFDM as an example of a multi-carrier scheme. 10 [0147] Fig. 13 shows the structure of a transmission device when using OFDM. In Fig. 13, elements that operate in a similar way to Fig. 3 bear the same reference signs. [0 1481 15 An OFDM related processor 1301A receives, as input, the weighted signal 309A, performs processing related to OFDM, and outputs a transmission signal 1302A. Similarly, an OFDM related processor 1301B receives, as input, the weighted signal 309B, performs processing related to OFDM, and outputs a transmission signal 1302B. 20 [0149] Fig. 14 shows an example of a structure from the OFDM related processors 1301A and 1301B in Fig. 13 onwards. The part fiom 1401A to 1410A is related to the part fkom 130 1 A to 3 12A in Fig. 13, and the part from 140 1 B to 14 10B is related to the part fiom 1301B to 312B in Fig. 13. 25 [0150] A seriaVparalle1 converter 1402A performs seriaVparalle1 conversion on a weighted signal 1401A (corresponding to the weighted signal 309A in Fig. 13) and outputs a parallel signal 1403A. 47 [0151] A reordering unit 1404A receives a parallel signal 1403A as input, performs reordering, and outputs a reordered signal 1405A. Reordering is described in detail later. 5 [0152] An inverse fast Fourier transformer 1406A receives the reordered signal 1405A as an input, performs a fast Fourier transform, and outputs a fast Fourier transformed signal 1407A. [O 1531 10 A wireless unit 1408A receives the fast Fourier transformed signal 1407A as an input, performs processing such as frequency conversion, amplification, and the like, and outputs a modulated signal 1409A. The modulated signal 1409A is output as a radio wave fkom an antenna 1410A. A seriaVparalle1 converter 1402B performs seriaVparalle1 conversion on a 15 weighted signal 1401B (corresponding to the weighted signal 309B in Fig. 13) and outputs a parallel signal 1403B. [0 1541 A reordering unit 1404B receives a parallel signal 1403B as input, performs reordering, and outputs a reordered signal 1405B. Reordering is described in detail 20 later. [0155] An inverse fast Fourier transformer 1406B receives the reordered signal 1405B as an input, performs a fast Fourier transform, and outputs a fast Fourier transformed signal 1407B. 25 [0156] A wireless unit 1408B receives the fast Fourier transformed signal 1407B as an input, performs processing such as frequency conversion, amplification, and the like, and outputs a modulated signal 1409B. The modulated signal 1409B is output as a radio wave from an antenna 141 0B. [0157] In the transmission device of Fig. 3, since the transmission scheme does not 5 use multi-carrier, precoding hops to form a four-slot period (cycle), as shown in Fig. 6, and the precoded symbols are arranged in the time domain. When using a multi-carrier transmission scheme as in the OFDM scheme shown in Fig. 13, it is of course possible to arrange the precoded symbols in the time domain as in Fig. 3 for each (sub)carrier. In the case of a multi-carrier transmission scheme, however, it is 10 possible to arrange symbols in the frequency domain, or in both the frequency and time domains. The following describes these arrangements. [0158] Figs. 15A and 15B show an example of a scheme of reordering symbols by reordering units 1401A and 1401B in Fig. 14, the horizontal axis representing 15 frequency, and the vertical axis representing time. The frequency domain runs fi-om (sub)carrier 0 through (sub)carrier 9. The modulated signals zl and 22 use the same fi-equency bandwidth at the same time. Fig. 15A shows the reordering scheme for symbols of the modulated signal zl, and Fig. 15B shows the reordering scheme for symbols of the modulated signal 22. Numbers #1, #2, #3, #4, . . . are assigned to in 20 order to the symbols of the weighted signal 1401A which is input into the serial/parallel converter 1402A. At this point, symbols are assigned regularly, as shown in Fig. 15A. The symbols #1, #2, #3, #4, ... are arranged in order starting fiom carrier 0. The symbols # 1 through #9 are assigned to time $1, and subsequently, the symbols # 10 through # 19 are assigned to time $2. 25 [0159] Similarly, numbers #1, #2, #3, #4, . . . are assigned in order to the symbols of the weighted signal 1401B which is input into the seriaVparalle1 converter 1402B. At this point, symbols are assigned regularly, as shown in Fig. 15B. The symbols #1, 49 #2, #3, #4, . . . are arranged in order starting fiom carrier 0. The symbols #I through #9 are assigned to time $1, and subsequently, the symbols #10 through #19 are assigned to time $2. Note that the modulated signals zl and 22 are complex signals. [0 1601 5 The symbol group 1501 and the symbol group 1502 shown in Figs. 15A and 15B are the symbols for one period (cycle) when using the precoding weight hopping scheme shown in Fig. 6. Symbol #O is the symbol when using the precoding weight of slot 4i in Fig. 6. Symbol #1 is the symbol when using the precoding weight of slot 4i + 1 in Fig. 6. Symbol #2 is the symbol when using the precoding 10 weight of slot 4i + 2 in Fig. 6. Symbol #3 is the symbol when using the precoding weight of slot 4i + 3 in Fig. 6. Accordingly, symbol #x is as follows. When x mod 4 is 0, the symbol #x is the symbol when using the precoding weight of slot 4i in Fig. 6. When x mod 4 is 1, the symbol #x is the symbol when using the precoding weight of slot 4i + 1 in Fig. 6. When x mod 4 is 2, the symbol #x is the symbol when using 15 the precoding weight of slot 4i + 2 in Fig. 6. When x mod 4 is 3, the symbol #x is the symbol when using the precoding weight of slot 4i + 3 in Fig. 6. [0161] In this way, when using a multi-carrier transmission scheme such as OFDM, unlike during single carrier transmission, symbols can be arranged in the frequency 20 domain. Furthermore, the ordering of symbols is not limited to the ordering shown in Figs. 15A and 15B. Other examples are described with reference to Figs. 16A, 16B, 17A, and 17B. [0 1 621 Figs. 16A and 16B show an example of a scheme of reordering symbols by 25 the reordering units 1404A and 1404B in Fig. 14, the horizontal axis representing fkequency, and the vertical axis representing time, that differs fkom Figs. 15A and 15B. Fig. 16A shows the reordering scheme for symbols of the modulated signal zl, and Fig. 16B shows the reordering scheme for symbols of the modulated signal 22. 5 0 The difference in Figs. 16A and 16B as compared to Figs. 15A and 15B is that the reordering scheme of the symbols of the modulated signal zl differs fiom the reordering scheme of the symbols of the modulated signal 22. In Fig. 16B, symbols #O through #5 are assigned to carriers 4 through 9, and symbols #6 through #9 are 5 assigned to carriers 0 through 3. Subsequently, symbols #10 through #19 are assigned regularly in the same way. At this point, as in Figs. 15A and 15B, the symbol group 160 1 and the symbol group 1 602 shown in Figs. 16A and 1 6B are the symbols for one period (cycle) when using the precoding weight hopping scheme shown in Fig. 6. 10 [0163] Figs. 17A and 17B show an example of a scheme of reordering symbols by the reordering units 1404A and 1404B in Fig. 14, the horizontal axis representing frequency, and the vertical axis representing time, that differs from Figs. 15A and 15B. Fig. 17A shows the reordering scheme for symbols of the modulated signal zl, 15 and Fig. 17B shows the reordering scheme for symbols of the modulated signal 22. The difference in Figs. 17A and 17B as compared to Figs. 15A and 15B is that whereas the symbols are arranged in order by carrier in Figs. 15A and 15B, the symbols are not arranged in order by carrier in Figs. 17A and 17B. It is obvious that, in Figs. 17A and 17B, the reordering scheme of the symbols of the modulated signal 20 zl may differ fiom the reordering scheme of the symbols of the modulated signal 22, as in Figs. 16A and 16B. [0 1641 Figs. 18A and 18B show an example of a scheme of reordering symbols by the reordering units 1404A and 1404B in Fig. 14, the horizontal axis representing 25 fiequency, and the vertical axis representing time, that differs fiom Figs. 15A through 17B. Fig. 18A shows the reordering scheme for symbols of the modulated signal zl, and Fig. 18B shows the reordering scheme for symbols of the modulated signal 22. In Figs. 15A through 17B, symbols are arranged in the frequency domain, whereas in Figs. 18A and 18B, symbols are arranged in both the frequency and time domains. [0 1651 In Fig. 6, an example has been described of hopping between precoding 5 weights over four slots. Here, however, an example of hopping over eight slots is described. The symbol groups 1801 and 1802 shown in Figs. 18A and 18B are the symbols for one period (cycle) when using the precoding weight hopping scheme (and are therefore eight-symbol groups). Symbol #O is the symbol when using the precoding weight of slot 8i. Symbol #I is the symbol when using the precoding 10 weight of slot 8i + 1. Symbol #2 is the symbol when using the precoding weight of slot 8i + 2. Symbol #3 is the symbol when using the precoding weight of slot 8i + 3. Symbol #4 is the symbol when using the precoding weight of slot 8i + 4. Symbol #5 is the symbol when using the precoding weight of slot 8i + 5. Symbol #6 is the symbol when using the precoding weight of slot 8i + 6. Symbol #7 is the symbol 15 when using the precoding weight of slot 8i + 7. Accordingly, symbol #x is as follows. When x mod 8 is 0, the symbol #x is the symbol when using the precoding weight of slot 8i. When x mod 8 is 1, the symbol #x is the symbol when using the precoding weight of slot 8i + 1. When x mod 8 is 2, the symbol #x is the symbol when using the precoding weight of slot 8i + 2. When x mod 8 is 3, the symbol #x is 20 the symbol when using the precoding weight of slot 8i + 3. When x mod 8 is 4, the symbol #x is the symbol when using the precoding weight of slot 8i + 4. When x mod 8 is 5, the symbol #x is the symbol when using the precoding weight of slot 8i + 5. When x mod 8 is 6, the symbol #x is the symbol when using the precoding weight of slot 8i + 6. When x mod 8 is 7, the symbol #x is the symbol when using 25 the precoding weight of slot 8i + 7. In the symbol ordering in Figs. 18A and 18B, four slots in the time domain and two slots in the frequency domain for a total of 4 x 2 = 8 slots are used to arrange symbols for one period (cycle). In this case, letting the number of symbols in one period (cycle) be m x n symbols (in other words, m x 52 n precoding weights exist), the number of slots (the number of carriers) in the frequency domain used to arrange symbols in one period (cycle) be n, and the number of slots used in the time domain be m, then m > n should be satisfied. This is because the phase of direct waves fluctuates more slowly in the time domain than in 5 the frequency domain. Therefore, since the precoding weights are changed in the present embodiment to minimize the influence of steady direct waves, it is preferable to reduce the fluctuation in direct waves in the period (cycle) for changing the precoding weights. Accordingly, m > n should be satisfied. Furthermore, considering the above points, rather than reordering symbols only in the frequency 10 domain or only in the time domain, direct waves are more likely to become stable when symbols are reordered in both the frequency and the time domains as in Figs. 18A and 18B, thereby making it easier to achieve the advantageous effects of the present invention. When symbols are ordered in the fiequency domain, however, fluctuations in the frequency domain are abrupt, leading to the possibility of yielding 15 diversity gain. Therefore, reordering in both the frequency and the time domains is not necessarily always the best scheme. [O 1661 Figs. 19A and 19B show an example of a scheme of reordering symbols by the reordering units 1404A and 1404B in Fig. 14, the horizontal axis representing 20 frequency, and the vertical axis representing time, that differs from Figs. 18A and 18B. Fig. 19A shows the reordering scheme for symbols of the modulated signal zl, and Fig. 19B shows the reordering scheme for symbols of the modulated signal 22. As in Figs. 18A and 18B, Figs. 19A and 19B show arrangement of symbols using both the frequency and the time axes. The difference as compared to Figs. 18A and 25 18B is that, whereas symbols are arranged first in the frequency domain and then in the time domain in Figs. 18A and 18B, symbols are ananged first in the time domain and then in the frequency domain in Figs. 19A and 19B. In Figs. 19A and 19B, the symbol group 1901 and the symbol group 1902 are the symbols for one period (cycle) when using the precoding hopping scheme. [O 1671 Note that in Figs. 18A, 18B, 19A, and 19B, as in Figs. 16A and 16B, the 5 present invention may be similarly embodied, and the advantageous effect of high reception quality achieved, with the symbol arranging scheme of the modulated signal zl differing from the symbol arranging scheme of the modulated signal 22. Furthermore, in Figs. 1 8A, 1 8B, 19A, and 19B, as in Figs. 1 7A and 17B, the present invention may be similarly embodied, and the advantageous effect of high reception 10 quality achieved, without arranging the symbols in order. [0168] Fig. 27 shows an example of a scheme of reordering symbols by the reordering units 1404A and 1404B in Fig. 14, the horizontal axis representing frequency, and the vertical axis representing time, that differs from the above 15 examples. The case of hopping between precoding matrices regularly over four slots, as in Equations 37-40, is considered. The characteristic feature of Fig. 27 is that symbols are arranged in order in the frequency domain, but when progressing in the time domain, symbols are cyclically shifted by n symbols (in the example in Fig. 27, n = I). In the four symbols shown in the symbol group 2710 in the frequency 20 domain in Fig. 27, precoding hops between the precoding matrices of Equations 3740. [0 1691 In this case, symbol #O is precoded using the precoding matrix in Equation 37, symbol #1 is precoded using the precoding matrix in Equation 38, symbol #2 is 25 precoded using the precoding matrix in Equation 39, and symbol #3 is precoded using the precoding matrix in Equation 40. [0 1701 Similarly, for the symbol group 2720 in the frequency domain, symbol #4 is precoded using the precoding matrix in Equation 37, symbol #5 is precoded using the precoding matrix in Equation 38, symbol #6 is precoded using the precoding matrix in Equation 39, and symbol #7 is precoded using the precoding matrix in 5 Equation 40. [0171] For the symbols at time $1, precoding hops between the above precoding matrices, but in the time domain, symbols are cyclically shifted. Therefore, precoding hops between precoding matrices for the symbol groups 2701,2702,2703, 10 and 2704 as follows. [0 1 721 In the symbol group 2701 in the time domain, symbol #O is precoded using the precoding matrix in Equation 37, symbol #9 is precoded using the precoding matrix in Equation 38, symbol #18 is precoded using the precoding matrix in 15 Equation 39, and symbol #27 is precoded using the precoding matrix in Equation 40. [0 1731 In the symbol group 2702 in the time domain, symbol #28 is precoded using the precoding matrix in Equation 37, symbol #1 is precoded using the precoding matrix in Equation 38, symbol #10 is precoded using the precoding matrix in 20 Equation 39, and symbol #19 is precoded using the precoding matrix in Equation 40. [0 1 741 In the symbol group 2703 in the time domain, symbol #20 is precoded using the precoding matrix in Equation 37, symbol #29 is precoded using the precoding matrix in Equation 38, symbol #2 is precoded using the precoding matrix in 25 Equation 39, and symbol # 11 is precoded using the precoding matrix in Equation 40. [0 1751 In the symbol group 2704 in the time domain, symbol #12 is precoded using the precoding matrix in Equation 37, symbol #21 is precoded using the precoding 5 5 matrix in Equation 38, symbol #30 is precoded using the precoding matrix in Equation 39, and symbol #3 is precoded using the precoding matrix in Equation 40. [0 1761 The characteristic of Fig. 27 is that, for example focusing on symbol #11, 5 the symbols on either side in the frequency domain at the same time (symbols #10 and #12) are both precoded with a different precoding matrix than symbol #11, and the symbols on either side in the time domain in the same carrier (symbols #2 and #20) are both precoded with a different precoding matrix than symbol #11. This is true not only for symbol #11. Any symbol having symbols on either side in the 10 frequency domain and the time domain is characterized in the same way as symbol #11. As a result, precoding matrices are effectively hopped between, and since the influence on stable conditions of direct waves is reduced, the possibility of improved reception quality of data increases. [0 1771 15 In Fig. 27, the case of n = 1 has been described, but n is not limited in this way. The present invention may be similarly embodied with n = 3. Furthermore, in Fig. 27, when symbols are arranged in the frequency domain and time progresses in the time domain, the above characteristic is achieved by cyclically shifting the number of the arranged symbol, but the above characteristic may also be achieved 20 by randomly (or regularly) arranging the symbols. [0178] (Embodiment 2) In Embodiment 1, regular hopping of the precoding weights as shown in Fig. 6 has been described. In the present embodiment, a scheme for designing specific 25 precoding weights that differ fiom the precoding weights in Fig. 6 is described. [0 1791 In Fig. 6, the scheme for hopping between the precoding weights in Equations 37-40 has been described. By generalizing this scheme, the precoding 56 weights may be changed as follows. (The hopping period (cycle) for the precoding weights has four slots, and Equations are listed similarly to Equations 37-40.) For symbol number 4i (where i is an integer greater than or equal to zero): [0 1 801 5 Math 42 Equation 42 [0181] 10 Here, j is an imaginary unit. For symbol number 4i + 1 : [0 1 821 Math 43 Equation 43 [0183] For symbol number 4i + 2: [0 1 841 20 Math 44 Equation 44 [0 1851 For symbol number 4i + 3: 5 [0186] Math 45 Equation 45 [0 1871 From Equations 36 and 41, the received vector R(t) = (rl(t), r2(t)lT can be represented as follows. For symbol number 4i: [0188] Math 46 Equation 46 [0189] 20 For symbol number 4i + 1 : [0 1901 Math 47 Equation 47 [0191] For symbol number 4i + 2: [0 1 921 Math 48 Equation 48 [0 1931 For symbol number 4i + 3: [0 1941 Math 49 Equation 49 [0 1951 In this case, it is assumed that only components of direct waves exist in the channel elements hll(t), h12(t), hZl(t), and hz2(t), that the amplitude components of the direct waves are all equal, and that fluctuations do not occur over time. With these assumptions, Equations 4649 can be represented as follows. For symbol number 4i: [0 1961 Math 50 Equation 50 [0 1971 5 For symbol number 4i + 1 : [0198] Math 51 Equation 5 1 10 [0 1991 I For symbol number 4i + 2: [0200] Math 52 15 Equation 52 [020 11 For symbol number 4i + 3: 20 [0202] Math 53 Equation 53 [0203] In Equations 50-53, let A be a positive real number and q be a complex 5 number. The values of A and q are determined in accordance with the positional relationship between the transmission device and the reception device. Equations 50-53 can be represented as follows. For symbol number 4i: 102041 10 Math 54 Equation 54 [0205] 15 For symbol number 4i + 1 : [0206] Math 55 Equation 55 [0207] For symbol number 4i + 2: [0208] Math 56 Equation 56 5 [0209] For symbol number 4i + 3: [02 1 01 Math 57 10 Equation 57 102 1 11 As a result, when q is represented as follows, a signal component based on 15 one of sl and s2 is no longer included in rl and r2, and therefore one of the signals sl and s2 can no longer be obtained. For symbol number 4i: [02 121 Math 58 20 Equation 58 For symbol number 4i + 1 : [02 141 Math 59 Equation 59 [02 1 51 For symbol number 4i + 2: [02 1 61 10 Math 60 Equation 60 102 1 71 15 For symbol number 4i + 3: [02 1 81 Math 61 Equation 6 1 20 [02 1 91 In this case, if q has the same solution in symbol numbers 4i, 4i + 1,4i + 2, and 4i + 3, then the channel elements of the direct waves do not greatly fluctuate. Therefore, a reception device having channel elements in which the value of q is 25 equivalent to the same solution can no longer obtain excellent reception quality for 63 any of the symbol numbers. Therefore, it is difficult to achieve the ability to correct errors, even if error correction codes are introduced. Accordingly, for q not to have the same solution, the following condition is necessary from Equations 58-61 when focusing on one of two solutions of q which does not include 6. 5 [0220] Math 62 Condition # 1 10 [0221] (xiso, 1,2,3;yisO, 1,2,3;andx#y.) In an example hlfilling Condition #1, values are set as follows: (Example # 1) (1) O11(4i) = O11(4i + 1) = O11(4i + 2) = O11(4i + 3) = 0 radians, 1 5 (2) 021(4i) = 0 radians, (3) 021(4i + 1) = x12 radians, (4) 021(4i + 2) = x radians, and (5) 021(4i + 3) = 3x12 radians. (The above is an example. It suffices for one each of zero radians, 1112 radians, n: 20 radians, and 3d2 radians to exist for the set (02,(4i), 021(4i + I), 021(4i + 2), 021(4i + 3)).) In this case, in particular under condition (I), there is no need to perform signal processing (rotation processing) on the baseband signal Sl (t), which therefore offers the advantage of a reduction in circuit size. Another example is to set values as follows. 25 (Example #2) (6) O1 l(4i) = 0 radians, (7) O1 l(4i + 1) = x/2 radians, (8) O1 ](4i + 2) = n radians, (9) O11(4i + 3) = 3d2 radians, and (10) 021(4i) = 021(4i + 1) = 02](4i + 2) = 02](4i + 3) = 0 radians. (The above is an example. It suffices for one each of zero radians, a12 radians, x radians, and 3d2 radians to exist for the set (el ](4i), OI1(4i + I), OI1(4i + 2), 01 1(4i + 3)).) In this case, in particular under condition (6), there is no need to perform signal processing (rotation processing) on the baseband signal S2(t), which therefore offers the advantage of a reduction in circuit size. Yet another example is as follows. (Example #3) (11) OI1(4i) =O11(4i + 1) +II1(4i + 2) +II1(4i + 3) = 0 radians, (12) 021(4i) = 0 radians, (13) 02](4i + 1) = x14 radians, (14) OZ1(4i + 2) = n12 radians, and (15) 02](4i + 3) = 3n14 radians. (The above is an example. It suffices for one each of zero radians, n14 radians, x12 radians, and 3x14 radians to exist for the set (02](4i), 02](4i + I), 021(4i + 2), 021(4i + 311.1 (Example #4) (1 6) O1 ](4i) = 0 radians, (17) O11(4i + 1) = n14 radians, (18) O11(4i + 2) = n12 radians, (19) O11(4i + 3) = 3x14 radians, and (20) 021(4i) = 02](4i + 1) = 02](4i + 2) = 021(4i + 3) = 0 radians. (The above is an example. It suffices for one each of zero radians, n14 radians, x12 radians, and 3d4 radians to exist for the set (OI1(4i), OI1(4i + I), OI1(4i + 2), O11(4i + 311.1 While four examples have been shown, the scheme of satisfling Condition #1 is not limited to these examples. 65 [0222] Next, design requirements for not only and el2, but also for h and 6 are described. It suffices to set h to a certain value; it is then necessary to establish requirements for 6. The following describes the design scheme for 6 when h is set to zero radians. [0223] In this case, by defining 6 so that n:12 radians 5 161 5 n: radians, excellent reception quality is achieved, particularly in an LOS environment. [0224] Incidentally, for each of the symbol numbers 4i, 4i + 1, 4i + 2, and 4i + 3, two points q exist where reception quality becomes poor. Therefore, a total of 2 x 4 = 8 such points exist. In an LOS environment, in order to prevent reception quality fiom degrading in a specific reception terminal, these eight points should each have a different solution. In this case, in addition to Condition #1, Condition #2 is necessary. [0225] Math 63 Condition #2 e j(6l 1(4i+xk621(4i+x)) + ej(611 (4i+~k62~(4i+f~ork VdX), V y (x,y = 0,1,2,3) and e l(*i+xk021(4i+xkd + ej(811 ( 4 i + y k 6 2 1 ( 4 i +t~6 ) for VX, vy (Xf y; X, y = 0,1,2,3) [0226] Additionally, the phase of these eight points should be evenly distributed (since the phase of a direct wave is considered to have a high probability of even distribution). The following describes the design scheme for 6 to satisfy this requirement. [0227] In the case of example #1 and example #2, the phase becomes even at the points at which reception quality is poor by setting 6 to * 3x14 radians. For example, letting 6 be 3d4 radians in example #1 (and letting A be a positive real number), then each of the four slots, points at which reception quality becomes poor exist 5 once, as shown in Fig. 20. In the case of example #3 and example #4, the phase becomes even at the points at which reception quality is poor by setting 6 to f x radians. For example, letting 6 be x radians in example #3, then in each of the four slots, points at which reception quality becomes poor exist once, as shown in Fig. 21. (If the element q in the channel matrix H exists at the points shown in Figs. 20 and 10 21, reception quality degrades.) With the above structure, excellent reception quality is achieved in an LOS environment. Above, an example of changing precoding weights in a four-slot period (cycle) is described, but below, changing precoding weights in an N-slot period (cycle) is described. Making the same considerations as in Embodiment 1 and 15 in the above description, processing represented as below is performed on each symbol number. For symbol number Ni (where i is an integer greater than or equal to zero): 102281 Math 64 20 Equation 62 [0229] Here, j is an imaginary unit. 25 For symbol number Ni + 1 : [0230] Math 65 Equation 63 5 [0231] When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - 1): [0232] Math 66 10 Equation 64 [023 31 Furthermore, for symbol number Ni + N - 1 : 15 [0234] Math 67 Equation 65 20 [0235] Accordingly, rl and r2 are represented as follows. For symbol number Ni (where i is an integer greater than or equal to zero): 68 [023 61 Math 68 Equation 66 5 [023 71 Here, j is an imaginary unit. For symbol number Ni + 1 : [023 81 10 Math 69 Equation 67 [023 91 15 When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - I): [0240] Math 70 Equation 68 1024 11 Furthermore, for symbol number Ni + N - 1 : [0242] 69 Math 71 Equation 69 5 [0243] In this case, it is assumed that only components of direct waves exist in the channel elements hll(t), h12(t), hzl(t), and hz2(t), that the amplitude components of the direct waves are all equal, and that fluctuations do not occur over time. With these assumptions, Equations 6649 can be represented as follows. 10 For symbol number Ni (where i is an integer greater than or equal to zero): [0244] Math 72 Equation 70 15 [0245] Here, j is an imaginary unit. For symbol number Ni + 1 : [0246] 20 Math 73 Equation 71 102471 When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - 1): [0248] 5 Math 74 Equation 72 [0249] 10 Furthermore, for symbol number Ni + N - 1 : [0250] Math 75 Equation 73 15 [025 11 In Equations 70-73, let A be a real number and q be a complex number. The values of A and q are determined in accordance with the positional relationship between the transmission device and the reception device. Equations 70-73 can be 20 represented as follows. For symbol number Ni (where i is an integer greater than or equal to zero): [0252] Math 76 Equation 74 [0253] Here, j is an imaginary unit. 5 For symbol number Ni + 1 : [0254] Math 77 Equation 75 10 [025 51 When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - 1): [025 61 15 Math 78 Equation 76 [025 71 20 Furthermore, for symbol number Ni + N - 1 : [025 81 Math 79 Equation 77 1025 91 5 As a result, when q is represented as follows, a signal component based on one of sl and s2 is no longer included in rl and r2, and therefore one of the signals sl and s2 can no longer be obtained. For symbol number Ni (where i is an integer greater than or equal to zero): [0260] 10 Math 80 Equation 78 [026 11 15 For symbol number Ni + 1 : [0262] Math 8 1 Equation 79 20 [0263] When generalized, this equation is as follows. ForsymbolnumberNi+k(k=O, 1, ..., N- 1): [0264] 25 Math 82 Equation 80 [0265] 5 Furthermore, for symbol number Ni + N - 1 : [0266] Math 83 Equation 8 1 - - A e j ( ~ l l ( ~ i + ~ - l t e 2 1 ( ~ i + ~ --l )A)e,j ( ~ ll ( ~ i + ~ - l t e 2 1 ( ~ i + ~ - l t s ) 10 [0267] In this case, if q has the same solution in symbol numbers Ni through Ni + N - 1, then since the channel elements of the direct waves do not greatly fluctuate, a reception device having channel elements in which the value of q is equivalent to 15 this same solution can no longer obtain excellent reception quality for any of the symbol numbers. Therefore, it is difficult to achieve the ability to correct errors, even if error correction codes are introduced. Accordingly, for q not to have the same solution, the following condition is necessary fiom Equations 78-81 when focusing on one of two solutions of q which does not include 6. 20 [0268] Math 84 Condition #3 $(el1 ( ~ l + x ) - e 2 1 ( ~ i + x ) )J. eI l ( ~ i + ~ t e 2 1 ( ~ l + ~fm) ) h ,y y (X y; = 0,1,2,.. . , N - 2,N - 1) Next, design requirements for not only and OI2, but also for h and 6 are described. It suff~cesto set h to a certain value; it is then necessary to establish requirements for 6. The following describes the design scheme for 6 when h is set to zero radians. 5 [0270] In this case, similar to the scheme of changing the precoding weights in a four-slot period (cycle), by defining 6 so that n:/2 radians 5 161 1 n: radians, excellent reception quality is achieved, particularly in an LOS environment. [027 11 10 In each symbol number Ni through Ni + N - 1, two points labeled q exist where reception quality becomes poor, and therefore 2N such points exist. In an LOS environment, in order to achieve excellent characteristics, these 2N points should each have a different solution. In this case, in addition to Condition #3, Condition #4 is necessary. 15 [0272] Math 85 Condition #4 eJ ~ +* eJ ~~~ ( M + Y+M , J M + ,Y ~,~fo)r vy ( ,y = 0,1,2,. .. , N - 2, N - I) and j b l J ~ ~ + x k ~ 2 1 ( ~ ~J ~+ ,~, (kN~~)+ Y M c " + Y ~ ~ ) f or yx, ~y (x + y; r,y = 0,1,2,. . ., N - 2, N - 1) 20 e * e [0273] Additionally, the phase of these 2N points should be evenly distributed (since the phase of a direct wave at each reception device is considered to have a high probability of even distribution). As described above, when a transmission device transmits a plurality of modulated signals from a plurality of antennas in a MIMO system, the advantageous effect of improved transmission quality, as compared to conventional spatial 75 multiplexing MIMO system, is achieved in an LOS environment in which direct waves dominate by hopping between precoding weights regularly over time. [0274] In the present embodiment, the structure of the reception device is as 5 described in Embodiment 1, and in particular with regards to the structure of the reception device, operations have been described for a limited number of antennas, but the present invention may be embodied in the same way even if the number of antennas increases. In other words, the number of antennas in the reception device does not affect the operations or advantageous effects of the present embodiment. 10 Furthermore, in the present embodiment, similar to Embodiment 1, the error correction codes are not limited. [0275] In the present embodiment, in contrast with Embodiment 1, the scheme of changing the precoding weights in the time domain has been described. As 15 described in Embodiment 1, however, the present invention may be similarly embodied by changing the precoding weights by using a multi-carrier transmission scheme and arranging symbols in the frequency domain and the frequency-time domain. Furthermore, in the present embodiment, symbols other than data symbols, such as pilot symbols (preamble, unique word, and the like), symbols for control 20 information, and the like, may be arranged in the frame in any way. [0276] (Embodiment 3) In Embodiment 1 and Embodiment 2, the scheme of regularly hopping between precoding weights has been described for the case where the amplitude of 25 each element in the precoding weight matrix is equivalent. In the present embodiment, however, an example that does not satisfl this condition is described. For the sake of contrast with Embodiment 2, the case of changing precoding weights over an N-slot period (cycle) is described. Making the same considerations 76 as in Embodiment 1 and Embodiment 2, processing represented as below is performed on each symbol number. Let P be a positive real number, and P # 1. For symbol number Ni (where i is an integer greater than or equal to zero): [0277] 5 Math 86 Equation 82 [0278] 10 Here, j is an imaginary unit. For symbol number Ni + 1 : [0279] Math 87 Equation 83 [0280] When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - 1): 20 [0281] Math 88 Equation 84 [0282] Furthermore, for symbol number Ni + N - 1 : 5 [0283] Math 89 Equation 85 10 [0284] Accordingly, rl and r2 are represented as follows. For symbol number Ni (where i is an integer greater than or equal to zero): [0285] Math 90 15 Equation 86 [0286] Here, j is an imaginary unit. 20 For symbol number Ni + 1 : [0287] Math 91 Equation 87 [028 81 5 When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - 1): [0289] Math 92 Equation 88 [0290] When generalized, this equation is as follows. For symbol number Ni + N - 1 : 15 [0291] Math 93 Equation 89 rl(Ni + N - 1) 1 4 (Ni + N - 1) h,, (Ni + N - I ) eJ61'N'+N-') I PxeJ ( B I ~ N * N - ~ ~S~l) ( ~+i N - 1) + N - 1) & (Ni + N - 1) p , (r-Z(Ni + N - 1)) = e~6z(N'+N-') ~b&'"+~-lb'~) sZ(Ni + N - 1) & [h: (N. e 1 20 [0292] In this case, it is assumed that only components of direct waves exist in the channel elements hll(t), hI2(t), h2](t), and hz2(t), that the amplitude components of the direct waves are all equal, and that fluctuations do not occur over time. With these assumptions, Equations 86-89 can be represented as follows. 25 For symbol number Ni (where i is an integer greater than or equal to zero): 79 * [0293] Math 94 Equation 90 5 [0294] Here, j is an imaginary unit. For symbol number Ni + 1 : [0295] 10 Math 95 Equation 91 [0296] 15 When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - I): [0297] Math 96 Equation 92 Furthermore, for symbol number Ni + N - 1 : [0299] Math 97 Equation 93 [0300] In Equations 90-93, let A be a real number and q be a complex number. Equations 90-93 can be represented as follows. 10 For symbol number Ni (where i is an integer greater than or equal to zero): [0301] Math 98 Equation 94 15 103021 Here, j is an imaginary unit. For symbol number Ni + 1 : [0303] 20 Math 99 Equation 95 [03 041 When generalized, this equation is as follows. For symbol number Ni + k (k = 0, 1, . . ., N - 1): [0305] 5 Math 100 Equation 96 [0306] 10 Furthermore, for symbol number Ni + N - 1 : [0307] Math 101 Equation 97 15 [0308] As a result, when q is represented as follows, one of the signals sl and s2 can no longer be obtained. For symbol number Ni (where i is an integer greater than or equal to zero): 20 [0309] Math 102 Equation 98 [03 lo] For symbol number Ni + 1 : [03 1 11 Math 103 5 Equation 99 A 4 = -ej ( o l , ( ~ j + l t o ~ ~ 0 ~ j-+ l A)B) , e j ( ~1l~ ~ i + l ) - ~ Z 1 0 V i + l t 6 ) P [03 121 When generalized, this equation is as follows. 10 For symbol number Ni + k (k = 0, 1, . . ., N - 1): [03 1 31 Math 104 Equation 100 A .( ~ j + k t e ~ ~ ( ~ i +-k A)P) ,e i(eII ( ~ i + k t e 2 1 ( ~ i + k t s ) , = --d e l l ( P 15 [03 141 Furthermore, for symbol number Ni + N - 1 : [03 1 51 Math 105 20 Equation 101 ~ z - A- ~ .J(@ 1 1 ( ~ i + ~ - l ) - e 2 1 ( ~ i + ~ --l ) A) P &(el1 ( ~ i + ~ - l ) - @ 2 1 ( ~ i + ~ - l ) 4 ) P Y [03 161 In this case, if q has the same solution in symbol numbers Ni through Ni + 25 N - 1, then since the channel elements of the direct waves do not greatly fluctuate, 83 excellent reception quality can no longer be obtained for any of the symbol numbers. Therefore, it is difficult to achieve the ability to correct errors, even if error correction codes are introduced. Accordingly, for q not to have the same solution, the following condition is necessary from Equations 98-101 when focusing on one 5 of two solutions of q which does not include 6. [03 1 71 Math 106 Condition #5 e ~ ( B i ~ ( ~ ~*~e~~ b ~k~6( ~~' ~~ y(k~6 ~~fo1r~( v~)x',)~v y y )ly(X) + ,,,; x,y = 0,1,2,. .. ,N - 2, N - 1) 10 [03 1 81 (xiso, 1,2, ..., N-2,N-l;yisO, l,2, ..., N-2,N-1;andxfy.) Next, design requirements for not only and eI2, but also for h and 6 are described. It suffices to set h to a certain value; it is then necessary to establish 15 requirements for 6. The following describes the design scheme for 6 when h is set to zero radians. [03 191 In this case, similar to the scheme of changing the precoding weights in a four-slot period (cycle), by defining 6 so that n12 radians 5 16) 5 n radians, excellent 20 reception quality is achieved, particularly in an LOS environment. [0320] In each of symbol numbers Ni through Ni + N - 1, two points q exist where reception quality becomes poor, and therefore 2N such points exist. In an LOS environment, in order to achieve excellent characteristics, these 2N points should 25 each have a different solution. In this case, in addition to Condition #5, considering that p is a positive real number, and j3 # 1, Condition #6 is necessary. [032 11 Math 107 Condition #6 e+ ++e~ b , , ( ~ ~ + y k % ~ ( ~ ' + r )f-oar )y x, yy (x y; x, y = 0,1,2,. .. , N - 2, N - I) [0322] 5 As described above, when a transmission device transmits a plurality of modulated signals from a plurality of antennas in a MIMO system, the advantageous effect of improved transmission quality, as compared to conventional spatial multiplexing MIMO system, is achieved in an LOS environment in which direct waves dominate by hopping between precoding weights regularly over time. 10 [0323] In the present embodiment, the structure of the reception device is as described in Embodiment 1, and in particular with regards to the structure of the reception device, operations have been described for a limited number of antennas, but the present invention may be embodied in the same way even if the number of 15 antennas increases. In other words, the number of antennas in the reception device does not affect the operations or advantageous effects of the present embodiment. Furthermore, in the present embodiment, similar to Embodiment 1, the error correction codes are not limited. 103241 20 In the present embodiment, in contrast with Embodiment 1, the scheme of changing the precoding weights in the time domain has been described. As described in Embodiment 1, however, the present invention may be similarly embodied by changing the precoding weights by using a multi-carrier transmission scheme and arranging symbols in the frequency domain and the frequency-time 25 domain. Furthermore, in the present embodiment, symbols other than data symbols, such as pilot symbols (preamble, unique word, and the like), symbols for control information, and the like, may be arranged in the frame in any way. [0325] 85 (Embodiment 4) In Embodiment 3, the scheme of regularly hopping between precoding weights has been described for the example of two types of amplitudes for each element in the precoding weight matrix, 1 and P. 5 [0326] In this case, [0327] Math 108 is ignored. [0329] Next, the example of changing the value of P by slot is described. For the 15 sake of contrast with Embodiment 3, the case of changing precoding weights over a 2 x N-slot period (cycle) is described. Making the same considerations as in Embodiment 1, Embodiment 2, and Embodiment 3, processing represented as below is performed on symbol numbers. Let p be a positive real number, and j3 # 1. Furthermore, let a be a positive real 20 number, and a # P. For symbol number 2Ni (where i is an integer greater than or equal to zero): [03 3 01 Math 109 Equation 102 [033 11 Here, j is an imaginary unit. 5 For symbol number 2Ni + 1 : [0332] Math 1 10 Equation 103 10 [0333] When generalized, this equation is as follows. For symbol number 2Ni + k (k = 0, 1, . . ., N - 1): LO3341 15 Math111 Equation 104 [0335] 20 Furthermore, for symbol number 2Ni + N - 1 : [0336] Math 1 12 Equation 105 [033 71 5 For symbol number 2Ni + N (where i is an integer greater than or equal to zero): [0338] Math 1 13 Equation 106 10 [0339] Here, j is an imaginary unit. For symbol number 2Ni + N + 1 : [0340] 15 Math114 Equation 107 [034 11 20 When generalized, this equation is as follows. Forsymbolnumber2Ni+N+k(k=O, 1, ..., N- I): [0342] Math 1 15 Equation 108 zl(2Ni + N + k) axej ( B , , ( Z ~ i + N + k ) + l))( sl(2Ni + N + k)) z2(2Ni + N + k) + + ej ( B , ( 2 M + ~ + k ) + l + ~ ) s 2 ( 2 ~+i N + k) [0343] 5 Furthermore, for symbol number 2Ni + 2N - 1 : [0344] Math 1 16 Equation 109 10 [0345] Accordingly, rl and r2 are represented as follows. For symbol number 2Ni (where i is an integer greater than or equal to zero): [0346] 15 Math117 Equation 1 10 [0347] 20 Here, j is an imaginary unit. For symbol number 2Ni + 1 : [0348] Math 1 18 Equation 1 1 1 [0349] When generalized, this equation is as follows. 5 For symbol number 2Ni + k (k = 0, 1, . . ., N - 1): [0350] Math 1 19 Equation 1 12 Furthermore, for symbol number 2Ni + N - 1 : [03 521 Math 120 15 Equation 1 13 For symbol number 2Ni + N (where i is an integer greater than or equal to zero): Math 121 Equation 1 14 rl(2Ni + N ) 1 h,, (2Ni + N) h,, (2Ni + (r2(2Ni + N))==(h,, (2Ni + N) h,(2Ni + [0355] Here, j is an imaginary unit. For symbol number 2Ni + N + 1 : [0356] 5 Math 122 Equation 1 15 10 When generalized, this equation is as follows. For symbol number 2Ni + N + k (k = 0, 1, . . ., N - 1): [03 5 81 Math 123 Equation 1 16 [03 5 91 When generalized, this equation is as follows. For symbol number 2Ni + 2N - 1 : 20 [0360] Math 124 Equation 1 17 In this case, it is assumed that only components of direct waves exist in the channel elements hll(t), h12(t), h2,(t), and h22(t), that the amplitude components of the direct waves are all equal, and that fluctuations do not occur over time. With these assumptions, Equations 1 10-1 17 can be represented as follows. 5 For symbol number 2Ni (where i is an integer greater than or equal to zero): [0362] Math 125 Equation 1 18 10 [0363] Here, j is an imaginary unit. For symbol number 2Ni + 1 : lo3641 15 Math126 Equation 1 19 [0365] 20 When generalized, this equation is as follows. For symbol number 2Ni + k (k = 0, 1, . . ., N - I): [0366] Math 127 Equation 120 [0367] Furthermore, for symbol number 2Ni + N - 1 : 5 [0368] Math 128 Equation 12 1 10 [0369] For symbol number 2Ni + N (where i is an integer greater than or equal to zero): [0370] Math 129 Equation 122 [0371] Here, j is an imaginary unit. For symbol number 2Ni + N + 1 : 20 [0372] Math 130 Equation 123 [0373] When generalized, this equation is as follows. 5 For symbol number 2Ni + N + k (k = 0, 1, . . ., N - 1): [0374] Math 13 1 Equation 124 10 [0375] Furthermore, for symbol number 2Ni + 2N - 1 : [0376] Math 132 15 Equation 125 [0377] In Equations 118-125, let A be a real number and q be a complex number. 20 Equations 1 18-1 25 can be represented as follows. For symbol number 2Ni (where i is an integer greater than or equal to zero): 103781 Math 133 Equation 126 [0379] Here, j is an imaginary unit. 5 For symbol number 2Ni + 1 : [0380] Math 134 Equation 127 10 [0381] When generalized, this equation is as follows. For symbol number 2Ni + k (k = 0, 1, . . ., N - I): [0382] 15 Math135 Equation 128 [0383] 20 Furthermore, for symbol number 2Ni + N - 1 : [03 841 ~ a t h1'3 6 Equation 129 [0385] For symbol number 2Ni + N (where i is an integer greater than or equal to zero): 5 [0386] Math 137 Equation 130 rl(2Ni + N) J @ , , ( ~ N ~ + N ) j ( B , , O ~ i + ~ ) t l ) axe sl(2Ni + N) (r2(2Ni + N)) = ~ [ ; : ) ( A ~ J ~ '( axe j o 2 , ( 2 ~ i + N ) J ( B ~ ~ ( ~ M + N ) + ~S+2~ ( 2 ~+i N ) e X 10 [0387] Here, j is an imaginary unit. For symbol number 2Ni + N + 1 : [0388] Math 1 3 8 15 Equation 13 1 When generalized, this equation is as follows. 20 For symbol number 2Ni + N + k (k = 0, 1, . . ., N - I): [0390] Math 139 Equation 132 [0391] Furthermore, for symbol number 2Ni + 2N - 1 : [0392] 5 Math 140 Equation 133 [0393] 10 As a result, when q is represented as follows, one of the signals sl and s2 can no longer be obtained. For symbol number 2Ni (where i is an integer greater than or equal to zero): [0394] Math 141 15 Equation 134 A ( 2" t g = e21 (2~j))-, AP e j ( ~ l l ( 2 ~ i ble(22 ~ i k s ) P [0395] For symbol number 2Ni + 1 : 20 [0396] Math 142 Equation 135 When generalized, this equation is as follows. For symbol number 2Ni + k (k = 0, 1, . . ., N - 1): [0398] Math 143 5 Equation 136 [0399] Furthermore, for symbol number 2Ni + N - 1 : 10 [0400] Math 144 Equation 137 15 [0401] For symbol number 2Ni + N (where i is an integer greater than or equal to zero): [0402] Math 145 Equation 138 [@031 For symbol number 2Ni + N + 1 : [0404] 25 Math 146 Equation 139 [0405] When generalized, this equation is as follows. Forsymbolnumber2Ni+N+k(k=O, 1, ..., N- I): [0406] Math 147 Equation 140 [04071 Furthermore, for symbol number 2Ni + 2N - 1 : [0408] 15 Math148 Equation 14 1 [04091 20 In this case, if q has the same solution in symbol numbers 2Ni through 2Ni + N - 1, then since the channel elements of the direct waves do not greatly fluctuate, excellent reception quality can no longer be obtained for any of the symbol numbers. Therefore, it is difficult to achieve the ability to correct errors, even if error correction codes are introduced. Accordingly, for q not to have the same solution, 25 Condition #7 or Condition #8 becomes necessary from Equations 134-141 and from 99 the fact that a # p when focusing on one of two solutions of q which does not include 6. [04101 Math 149 5 Condition #7 ei (Bl1(2~i+*be21(2+~ &;+(B~l )I)( ~ N ; + Y ~ O Z ~ ( ~ ~ ~fo+rY V)X), VY (X # y; x,y = 0,1,2,.. . , N - 2, N - 1) (xiso, 1,2, ..., N-2,N-l;yisO, 1,2, ..., N-2,N- 1;andxfy.) and e ~ ( B 1 1 ( 2 ~ J + ~ + x k ~ 2 1 ( 2 ~ ~,+~~(+Bx~)~)(~~ N'+N+Y~B~~(Z~+Nf+orY )V)x ,Vy (x f y; x, y = 0,1,2;.., N -2, N -1) 10 (xiso, 1,2, ..., N-2,N- l;yisO, 1,2, ..., N-2,N- 1;andxfy.) 11 Math 150 Condition #8 15 ,i(B1,(2~~+xte,,(2~i+x))e#i b11(2~+~te21(2fo~r V;+X~,v y) )(X # y; X,y = 0,1,2,.. . ,2N - 2,2N -1) [04 1 21 In this case, Condition #8 is similar to the conditions described in I Embodiment 1 through Embodiment 3. However, with regards to Condition #7, 20 since a # P, the solution not including 6 among the two solutions of q is a different solution. [04 1 31 Next, design requirements for not only ell and €Il2, but also for h and 6 are described. It suffices to set h to a certain value; it is then necessary to establish 25 requirements for 6. The following describes the design scheme for 6 when h is set to zero radians. CLAIMS 1. A precoding method for generating, from a plurality of baseband signals, a plurality of precoded signals that are transmitted in the same frequency bandwidth at 5 the same time, the precoding method comprising the steps of: selecting one matrix from among 2N matrices F[i], wherein i = 0, 1, 2, . . ., 2N-2, 2N-1, by hopping between the matrices, the 2N matrices F[i] defining a precoding process that is performed on the plurality of baseband signals; multiplying "u" by a first baseband signal sl generated from a first set of 10 bits, multiplying "v" by a second baseband signal s2 generated from a second set of bits, "u" and "v" denoting real numbers different from each other; and generating a first precoded signal zl and a second precoded signal 22 by performing a precoding process, which corresponds to a matrix selected fi-om among the 2N matrices F[i], on a signal obtained by multiplying "u" by the first baseband 15 signal sl and a signal obtained by multiplying "v" by the second baseband signal s2, the first precoded signal zl and the second precoded signal 22 satisfling (zl, ~ 2=) F[~i] ( u x sl, v x s21T, for i = 0, 1,2, . . ., N-2, N-1, the 2N matrices F[i] being expressed as: Math 1 20 Equation 279 for i = N, N+l, N+2, ..., 2N-2, 2N-1, the 2N matrices F[i] being expressed as: Math 2 25 Equation 280 h representing an arbitrary angle, a representing a positive real number excluding 1, el and eZl(i) satisfying: Math 3 5 Condition #57 ej (ellb)-e21G)t) e~(811(Y)-e21Q) for VX,v y ((x t y; X,y = 0.1,2,-. ., N - 2. N - 1) and Math 4 Condition #62 10 e~(e1i( ~ke21t( ~e~))( ellwe21(J')) fix a ,a y (x t y; x, y = N,N + 1. N + 2,..-.2N- 2,2N - 1) each of the 2N matrices being selected at least once in a predetermined time period. 2. A precoding apparatus for generating, from a plurality of baseband signals, a 15 plurality of precoded signals that are transmitted in the same frequency bandwidth at the same time, the precoding apparatus comprising: a weighting information generating unit selecting one matrix from among 2N matrices F[i], wherein i = 0, 1, 2, . . ., 2N-2, 2N-1, by hopping between the matrices, the 2N matrices F[i] defining a precoding process that is performed on the 20 plurality of baseband signals; a power changing unit multiplying "u" by a first baseband signal sl generated from a first set of bits, multiplying "v" by a second baseband signal s2 generated from a second set of bits, "u" and "v" denoting real numbers different &om each other; and 25 a weighting unit generating a first precoded signal zl and a second precoded signal 22 by performing a precoding process, which corresponds to a matrix selected !Q fiom among the 2N matrices F[i], on a signal obtained by multiplying "u" by the first baseband signal sl and a signal obtained by multiplying "v" by the second baseband signal s2, the first precoded signal zl and the second precoded signal 22 satisfying (21, for i = 0, 1,2, . . ., N-2, N-1, the 2N matrices F[i] being expressed as: Math 5 Equation 279 - 10 for i = N, N+l, N+2, .. ., 2N-2, 2N-1, the 2N matrices F[i] being expressed as: Math 6 Equation 280 F[il = 1 -("" is1 lid e j ( ~l(li) +") ) ,/,"+I $821(i) a x ,~(e2l(ik*+~) 15 h representing an arbitrary angle, a representing a positive real number excluding 1, O1 l(i) and eZl(i) satisfling: Math 7 Condition #57 e ~ ( ~ ~ ~+( e~~(e~k1(Y8)-0~21(~Y)) ( f~or V)X, )VY (X+ Y; X,Y= 0,1,2,.--,N- 2, N -1) 20 and Math 8 Condition #62 e ~ (1~(~i k 821t(~ $)()B l l b k h l ( ~ ) ) for Kc, Vy (x f y; x, y = N,N+l,N + 2;..,2N - 2,2N -1) each of the 2N matrices being selected at least once in a predetermined time 25 period. 648