DESC:TECHNICAL FIELD
The present invention relates to satellite interception, especially VSAT (Very Small Aperture Terminal) monitoring and analysis. The present invention particularly relates to blind demodulation of Carrier-in-Carrier (CiC) signals and Paired Carrier Multiple Access (PCMA) signals wherein the outbound and inbound signals occupy the same frequency band.

BACKGROUND
The present invention relates to satellite communication signal interception technology, especially VSAT (Very Small Aperture Terminal) interception. It is a technical challenge to perform blind demodulation of the Carrier-in-Carrier (CiC) signals and Paired Carrier Multiple Access (PCMA) signals wherein the outbound and inbound signals occupy the same frequency band.
In patent publication CN104320362A titled, “Method for PCMA signal blind separation”, a method for PCMA signal separation is disclosed. The method comprises of: obtaining a plurality of initial particles according to distribution of initial prior probabilities of channel parameters and two paths of symbols; updating the channel parameters of the particles at the current moment; selection of available particles depending on the posterior probability values; updating and normalizing weights of the available particles; updating the two paths of symbols corresponding to the available particles; deciding the number of effective particles at the current moment; and estimating the channel parameters and the two paths of symbols of the PCMA signals according to the weights of the available particles when the number of the effective particles reaches a pre-set threshold value.
In the patent CN109412725B titled, “Radio communication PCMA signal blind demodulation method and device”, a PCMA signal blind demodulation method is disclosed. The method comprises: training and learning the neural network taking the characteristic data (extracted data containing frequency information and energy distribution information of PCMA signals constructed using channel parameters and different bit sequences) as input and the bit sequence as output; extracting characteristic data of the received PCMA signal in the radio communication process; feeding the characteristic data into the trained neural network; and outputting the demodulation result of the received PCMA signal as a bit sequence.
In the patent CN111526103B titled “PCMA signal single-channel blind separation method”, a single-channel blind separation methods based on a genetic improved particle filter algorithm is disclosed. The method comprises of: performing estimation on the signals and the channel parameters; generating a prediction signal according to the estimation parameters and the generated prediction symbol sequence by establishing a plurality of state distributions; obtaining a particle probability value corresponding to each particle; introducing selective cross operation in a genetic algorithm to generate new particles; sorting all the current particles according to the evaluation values to generate preferred particles; continuously optimizing particles by adding a dichotomy and outputting a separated symbol sequence.
In the patent publication KR20180032522A titled “Joint identification of merge signals in non-cooperative digital telecommunications”, a method for real-time blind separation and demodulation of a digital communication signal from a composite signal observation is disclosed. The method comprises of pre-processing the sum signal; estimating a parameter of a channel in a maximum likelihood through an adapted EM algorithm, wherein the log-likelihood conditional expectation value calculation is performed recursively by a particle filtering smoothing method; jointly demodulating channels according to a stochastic Viterbi algorithm; and tracking the temporal evolution of each channel parameter.
Paper titled “Single-Channel Blind Separation of Co-Frequency PSK Signals with Unknown Carrier Frequency Offsets” investigates the problem of single-channel blind source separation (SCBSS) of a mixture of two co-frequency phase-shift keying (PSK) signals with unknown carrier frequency offsets (CFOs). The paper proposes two algorithms that use the conventional Per Survivor Processing (PSP) algorithm to perform separation of the mixture signals. The phase rotations of the received signals are tracked by estimating the phase changes in each symbol interval.
Paper titled, “The DFF-PSP Iterative Separation and Theoretical Bound for PCMA with Long Memory” proposes the Decision feedback-feedforward per-survivor processing (DFF-PSP) iterative separation algorithm for Paired carrier multiple access (PCMA) Single channel blind separation (SCBS). This algorithm is overly complex with long memory. The truncated PSP is employed while the pre- and post- cursors are disregarded to control the complexity. The delayed decisions feed forward, and local decisions feedback filters are designed to process the pre- and post-cursors. The Bit-interleaved code modulation iterative decoder (BICM-ID) is combined with the second soft decision of the truncated PSP to obtain a better performance.
The drawback of most of the methods is that the real time separation of the mixed signals is difficult as they are computationally intensive. Also, most of the methods assume that the carrier frequency offsets of the mixed signals are known at the receiver or assume that there is no carrier frequency offset.
The conventional/traditional VSAT (Very Small Aperture Terminal) monitoring equipment used for interception works very well if the outbound and inbound signals use different frequency bands. In order to optimize bandwidth utilization in VSAT communication networks/links, Carrier-in-Carrier (CiC) technology is used by overlapping outbound and inbound signals of a VSAT network. The conventional VSAT monitoring/intercepting equipment cannot demodulate and decode the CiC signals wherein both the outbound and inbound links are transmitted at the same frequency and share same frequency band in case of duplex SCPC (Single Channel Per Carrier) links.
In such cases, separating the combined CiC signals (wherein two M-PSK signals are mixed in the same frequency band) is an essential pre-condition for the successful interception of a wide range of satellite communication links.
The problem of single channel blind source separation of M-PSK signals is difficult to solve as two source signals need to be extracted from one single observation. Accordingly, there is a need for a system and method for CiC signal separation, more specifically M-PSK signals.

SUMMARY OF THE INVENTION
This summary is provided to introduce concepts of the present invention. This summary is neither intended to identify essential features of the present invention nor is it intended for use in determining or limiting the scope of the present invention.
In accordance with the present invention, a system for single channel blind source separation of M-Phase Shift Keying (M-PSK) signals in Very Small Aperture Terminal (VSAT) interception is disclosed. The system comprises: an antenna sub-system configured to receive the signals from the satellite transponder; a front end RF system having a Low Noise Amplifier (LNA) configured to amplify the received signals, a down converter configured to down convert the signal and an Intermediate-frequency (IF) Amplifier configured to reduce noise and intercept the signal efficiently; a separator hardware configured to receive the composite signal, wherein the composite signal comprises, two M-PSK signals mixed in the same frequency band, and the separator hardware is further configured to output two separated M-PSK signals occupying different frequency bands; a Radio Frequency (RF) transceiver comprising an Analog to Digital Converter (ADC) configured to capture the composite data; a Field Programmable Gate Array (FPGA) board comprising a pre-processing unit configured to filter the captured data using root raised-cosine filter and to perform symbol timing synchronization; a processor based hardware comprising a Carrier Frequency Offset (CFO) estimation unit configured to estimate two carrier frequency offsets corresponding to the two M-PSK signals from the symbol synchronized composite data; the processor based hardware comprising a signal separation unit configured to estimate the transmitted symbol sequences corresponding to the two M-PSK signals by incorporating the pre-estimated CFO information and by estimating the channel impulse responses of the two channels involved using the symbol synchronized composite data; the FPGA board comprising a post-processing unit configured to re-modulate the estimated symbol sequences to M-PSK signals centered at two different frequencies as opposed to the composite signal wherein the M-PSK signals are centered at the same frequency; and the RF transceiver comprising a Digital to Analog Converter (DAC) configured to convert the digital M-PSK waveforms to analog M-PSK signals.
In one aspect, the separator hardware, comprises: the Carrier Frequency Offset (CFO) estimation unit configured to estimate the carrier frequency offsets of the two mixing signals before the start of real time signal separation, by maintaining a larger set of survivors at each symbol; and, the signal separation unit configured to perform real time composite signal separation by incorporating the estimated CFOs from the CFO estimation unit and by maintaining a smaller set of survivors at each symbol.
In another aspect, the CFO estimation unit comprises of: a state table creation unit configured to generate all the possible combinations of the symbols belonging to the two mixing M-PSK signals, a branch metric computation unit configured to compute the branch metric incurred for a survivor to enter a state, an accumulated metric computation unit configured to compute the accumulated metric incurred for a survivor to enter a state, wherein the accumulated metric includes summation of all the branch metrics in the surviving path, a survivors’ selection unit configured to select the best survivors from all the possible survivors, a Channel Impulse Response (CIR) updation unit configured to update the channel impulse response coefficients for all the survivors; and a phase rotation estimation and updation unit to update the CFOs by estimating the phase rotations for all the survivors.
In yet another aspect, the signal separation unit comprises: a state table creation unit configured to generate all the possible combinations of the symbols belonging to the two mixing M-PSK signals; a branch metric computation unit configured to compute the branch metric incurred for a survivor to enter a state; an accumulated metric computation unit configured to compute the accumulated metric incurred for a survivor to enter a state, wherein the accumulated metric includes summation of all the branch metrics in the surviving path; a survivors’ selection unit configured to select the best survivors from all the possible survivors; a Channel Impulse Response (CIR) updation unit configured to update the channel impulse response coefficients for all the survivors; and a trace back unit configured to output the estimated maximum likelihood source symbol sequences or state sequences.
In another aspect of the invention, a method for single channel blind source separation of M-Phase Shift Keying (M-PSK) signals in Very Small Aperture Terminal (VSAT) interception is disclosed. The method comprises: creating a state table comprising all the possible combinations of the symbols belonging to the two mixing M-PSK signals; receiving a composite signal, at a data acquisition unit, to process the composite signal; filtering the received composite data, using a root raised cosine (RRC) filter, to reduce noise; processing the received filtered composite data and output time synchronized symbols at symbol rate; computing branch metric incurred for each survivor to enter a state at each time synchronized symbol; computing accumulated metric incurred for each survivor to enter a state at each time synchronized symbol; selecting one or more survivors by sorting the accumulated metrics; updating Channel Impulse Response (CIR) coefficients for all the survivors based on the branch error incurred; estimating phase rotation and updating phase rotation for all the survivors and updating the Carrier Frequency Offsets (CFOs); and outputting estimated symbols by tracing back the survivor with least accumulated metric by a length specified as Trace Back Length (TBL).
In another aspect, computing branch metric comprises: including the pre-cursor Inter Symbol Interference (ISI) and post-cursor Inter Symbol Interference (ISI); computing branch metric from each survivor to a specific state as the sum of the squares of absolute values of a first error component and a second error component, wherein: the first error component corresponds to the difference between the symbol timing synchronized composite data and the post cursor ISI components along with the current symbol components; the second error component corresponds to the difference between the delayed first error component and the pre cursor ISI components.
In another aspect, selecting one or more survivors comprises: using two levels of sorting for reduction in execution time to support real time composite signal separation without affecting the separation capability, comprising: computation of minimum accumulated metric of a state as the least of the accumulated metrics from all the survivors to a state; the first level of sorting for the selection of states by sorting the minimum accumulated metrics of the states; and the second level of sorting for the selection of survivors from the selected states by sorting the accumulated metrics.
In another aspect, updating Channel Impulse Response (CIR) coefficients for all the survivors based on the branch error incurred comprises using the first error component for updating the CIR coefficients corresponding to the current symbol and post cursor ISI components using least mean squares algorithm; and using the second error component for updating the CIR coefficients corresponding to the pre cursor ISI components using least mean squares algorithm.
In another aspect, estimating phase rotation and updating phase rotation for the survivors and updating the Carrier Frequency Offsets (CFOs) comprises estimating two carrier frequency offsets corresponding to the two M-PSK signals by estimating the phase rotations and passing the estimated phase rotations through PLL (Phase Locked Loop).

BRIEF DESCRIPTION OF ACCOMPANYING DRAWINGS
The detailed description is described with reference to the accompanying figures. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. The same numbers are used throughout the drawings to reference like features and modules.
Figure 1 illustrates a system for single channel blind source separation of M-PSK (M-ary Phase Shift Keying) Signals in VSAT (Very Small Aperture Terminal) interception, according to an exemplary implementation of the present invention.
Figure 2 illustrates a state table used for separating a composite signal obtained by mixing two QPSK signals, according to an exemplary implementation of the present invention.
Figure 3 illustrates a carrier frequency offsets (CFO) estimation unit, according to an exemplary implementation of the present invention.
Figure 4 illustrates various blocks/units in a signal separation unit, according to an exemplary implementation of the present invention.
Figure 5 illustrates a method for single channel blind source separation, according to an exemplary implementation of the present invention.
Figure 6 illustrates a representation of the branch metric computation from the survivors to the states, according to an exemplary implementation of the present invention.
It should be appreciated by those skilled in the art that any block diagrams herein represent conceptual views of illustrative methods embodying the principles of the present invention. Similarly, it will be appreciated that any flow charts, flow diagrams, and the like represent various processes which may be represented in computer readable medium and so executed by a computer or processor, whether or not such computer or processor is explicitly shown.

DETAILED DESCRIPTION
The various embodiments of the present invention describe about blind demodulation of Carrier-in-Carrier (CiC) signals and Paired Carrier Multiple Access (PCMA) signals wherein the outbound and inbound signals occupy the same frequency band.
This invention discloses a system and method for single channel blind source separation of M-PSK Signals in VSAT interception wherein a blind and joint estimation of maximum likelihood transmitted source symbol sequences, channel impulse response coefficients and carrier frequency offsets corresponding to the two mixing M-PSK signals using per-survivor processing is performed. Per-Survivor Processing (PSP) technique is used for approximating Maximum Likelihood Sequence Estimation (MLSE) for unknown channels by embedding channel estimation into the structure of the Viterbi Algorithm (VA).
In the following description, for purpose of explanation, specific details are set forth in order to provide an understanding of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced without these details. One skilled in the art will recognize that embodiments of the present invention, some of which are described below, may be incorporated into a number of systems.
However, the systems and methods are not limited to the specific embodiments described herein. Further, structures and devices shown in the figures are illustrative of exemplary embodiments of the present invention and are meant to avoid obscuring of the present invention.
It should be noted that the description merely illustrates the principles of the present invention. It will thus be appreciated that those skilled in the art will be able to devise various arrangements that, although not explicitly described herein, embody the principles of the present invention. Furthermore, all examples recited herein are principally intended expressly to be only for explanatory purposes to help the reader in understanding the principles of the invention and the concepts contributed by the inventor to furthering the art and are to be construed as being without limitation to such specifically recited examples and conditions. Moreover, all statements herein reciting principles, aspects, and embodiments of the invention, as well as specific examples thereof, are intended to encompass equivalents thereof.
In one embodiment, a system for single channel blind source separation of M-Phase Shift Keying (M-PSK) signals mixed in the same frequency band in Very Small Aperture Terminal (VSAT) interception is disclosed. The system comprises: an antenna sub-system configured to receive the signals from the satellite transponder, a front end RF system having a Low Noise Amplifier (LNA) configured to amplify the received signals, a down converter configured to down convert the signal and an Intermediate-frequency (IF) Amplifier configured to reduce noise and intercept the signal efficiently, a separator hardware configured to receive the composite signal and output two separated M-PSK signals occupying different frequency bands, a Radio Frequency (RF) transceiver comprising an Analog to Digital Converter (ADC) configured to capture the composite data, a Field Programmable Gate Array (FPGA) board comprising a pre-processing unit configured to filter the captured data using root raised-cosine filter and to perform symbol timing synchronization; a processor based hardware comprising a Carrier Frequency Offset (CFO) estimation unit configured to estimate two carrier frequency offsets corresponding to the two M-PSK signals from the symbol synchronized composite data; the processor based hardware comprising a signal separation unit configured to estimate the transmitted symbol sequences corresponding to the two M-PSK signals by incorporating the pre-estimated CFO information and by estimating the channel impulse responses of the two channels involved using the symbol synchronized composite data.
In one embodiment, the separator hardware comprises of: a Carrier Frequency Offset (CFO) estimation unit configured to estimate the carrier frequency offsets of the two mixing signals before the start of real time signal separation, by maintaining a larger set of survivors at each symbol; and, a signal separation unit configured to perform real time composite signal separation by incorporating the estimated CFOs from the CFO estimation unit and by maintaining a smaller set of survivors at each symbol.
In one embodiment, a method to perform a blind and joint estimation of the unknown parameters such as maximum likelihood transmitted source symbol sequences, the channel impulse response coefficients and the carrier frequency offsets (CFOs) corresponding to the two mixing M-PSK signals is disclosed. The method includes the compensation of the pre-cursor and the post-cursor ISI (Inter Symbol Interference) components. The method includes usage of Phase Locked Loop (PLL) for accurate estimation of the CFOs of the mixing signals. The disclosed system performs pre-processing of the composite signal in the Field Programmable Gate Array (FPGA), actual signal separation in the processor based hardware and post-processing (wherein the estimated maximum likelihood source symbol sequences are re-modulated to QPSK signals centred at two different frequencies) in the Field Programmable Gate Array (FPGA).
Figure 1 illustrates a system for single channel blind source separation of M-PSK (M-ary Phase Shift Keying) Signals in VSAT (Very Small Aperture Terminal) interception. The system comprises an antenna sub-system (104), a RF front end system (106), a separator hardware (120), a Radio Frequency (RF) transceiver (124), a Field Programmable Gate Array (FPGA) board (128) and a processor based hardware (132). The antenna sub-system 104 receives the VSAT signals from the satellite transponder 102. The RF front end system 106 intercepts the received signal efficiently, and has an LNA (Low Noise Amplifier) 108 for amplifying the signal with minimal noise, a down converter 110 and IF Amplifier 118 for amplifying the IF signal. The down converter 110 consists of a Microwave oscillator 112, a Mixer 114 and a BPF (Band Pass Filter) 116 to down convert the signal from higher frequency to lower frequency.
The separator hardware 120 takes the type of modulation of the mixing M-PSK signals (such as BPSK, QPSK, 8-PSK) and the symbol rate of the mixing signals as inputs (the symbol rates of both the mixing signals are assumed to be the same). The down converted, amplified and noise reduced signal from the RF front end system 106 is fed to the separator hardware 120 which is capable of receiving the composite signal and outputs two separated M-PSK signals occupying different frequency bands. In one embodiment, the two M-PSK signals are mixed in the same frequency band.
The separator hardware 120 performs blind signal separation as it has no other information regarding either the mixing signals in the composite signal or the header formats of the mixing signals and finally outputs the M-PSK signals with the same symbol rate as that of the input. The RF transceiver 124 comprises of an Analog to Digital Converter (ADC) 126 that captures the composite data at the required sampling rate (=4*symbol rate) and passes to a Field Programmable Gate Array (FPGA) board 128. The FPGA board 128 contains a pre-processing unit 130 that filters the digitized captured data using RRC (root raised-cosine) filter and performs symbol timing synchronization. The filtered composite symbols are passed to the processor based hardware 132 using Ethernet.
The CFO (Carrier Frequency Offset) estimation unit 134 in the processor based hardware 132 estimates the two carrier frequency offsets corresponding to the two M-PSK signals. The signal separation unit 136 in the processor based hardware 132 estimates the transmitted symbol sequences (sequences with maximum likelihood) corresponding to the two M-PSK signals by incorporating the pre-estimated CFO information from the CFO estimation unit 134 and by estimating the channel impulse responses of the two channels involved using the symbol synchronized composite data. In other words, the signal separation unit 136 performs joint detection of transmitted symbol sequences and the channel impulse responses corresponding to the two mixing M-PSK signals. The estimated maximum likelihood symbol sequences (or bit sequences) are communicated from the processor based hardware 132 to the FPGA board 128 through Ethernet.
The post-processing unit 138 on the FPGA board 128 re-modulates the estimated symbol sequences (or bit sequences) to M-PSK signals centered at two different frequencies as opposed to the composite signal wherein the M-PSK signals are centered at the same frequency hindering the conventional demodulation and decoding. The DAC (Digital to Analog Converter) 140 in the RF transceiver 124 converts the digital M-PSK waveforms to analog M-PSK signals. The separated M-PSK waveforms are passed to the conventional demodulation and decoding facility 142 to demodulate and decode the actual transmitted data (after applying the correct FEC (Forward Error Correction) decoders, de-scramblers, de-framers, decompression, voice decoders, etc.). In cases where a direct conventional decoding facility is available, the demodulation of the separated bit sequences can be by-passed without performing the post-processing, DAC conversion and demodulation (i.e.), the decoding facility 144 decodes the estimated bit sequences directly.
As mentioned earlier, the composite signal separation is performed by joint detection of transmitted symbol sequences and the channel impulse responses corresponding to the two mixing M-PSK signals. Therefore, in order to perform the joint detection, a table referred to as the state table is created consisting of all the possible combinations of the transmitted source symbols depending on the modulation scheme employed.
Figure 2 illustrates a state table used for separating a composite signal obtained by mixing two QPSK signals. If a signal is QPSK modulated, bits are mapped to one of the four symbols represented as 0.707+0.707j,0.707-0.707j,-0.707-0.707j and -0.707+0.707j. Therefore, when two QPSK signals are mixed, each signal can take one of the four symbols mentioned above giving rise to 16 combinations as shown in the state table of Figure 2. Each combination is referred to as a state and is represented using a state index. For an M-PSK signal, M2 states are present.
Figure 3 illustrates a carrier frequency offsets (CFO) estimation unit. The CFO estimation unit 134 in the separator hardware 120 comprises of state table creation unit 302 configured to generate all the possible combinations of the symbols belonging to the two mixed M-PSK signals wherein each combination represents a state; a branch metric computation unit 304 configured to compute the branch metric incurred for a survivor to enter a state; accumulated metric computation unit 306 configured to compute the accumulated metric incurred for a survivor to enter a state wherein the accumulated metric includes summation of all the branch metrics in the surviving path; a survivors’ selection unit 308 configured to select the best survivors from all the possible survivors; a Channel Impulse Response (CIR) updation unit 310 configured to update the channel impulse response coefficients for all the survivors; and phase rotation estimation and updation unit 312 to update the carrier frequency offsets by estimating the phase rotations for all the survivors.
Figure 4 illustrates various blocks/units in a signal separation unit. The signal separation unit 136 in the separator hardware 120 comprises of: a state table creation unit 402 configured to generate all the possible combinations of the symbols belonging to the two mixing M-PSK signals where each combination represents a state; a branch metric computation unit 404 configured to compute the branch metric incurred for a survivor to enter a state; an accumulated metric computation unit 406 configured to compute the accumulated metric incurred for a survivor to enter a state wherein the accumulated metric includes summation of all the branch metrics in the surviving path; a survivors’ selection unit 408 configured to select the best survivors from all the possible survivors; a Channel Impulse Response (CIR) updation unit 410 configured to update the channel impulse response coefficients for all the survivors; and a trace back unit 412 configured to output the estimated maximum likelihood source symbol sequences or state sequences.
Figure 5 illustrates a method for single channel blind source separation. The method comprises of the following steps of: creating a state table (502) comprising all the possible combinations of the symbols belonging to the two mixing M-PSK signals; receiving (504) a composite signal, at a data acquisition unit (504), to process the composite signal and convert to composite data; filtering (506) the data, using a root raised cosine (RRC) filter, to reduce noise; processing (508) the received filtered composite data and output time synchronized symbols at symbol rate; computing (510) branch metric incurred for each survivor to enter a state at each time synchronized symbol; computing (512) accumulated metric incurred for each survivor to enter a state at each time synchronized symbol; selecting (514) one or more survivors by sorting the accumulated metrics; updating (516) Channel Impulse Response (CIR) coefficients for all the survivors based on the branch error incurred; estimating (518) phase rotation and updating phase rotation for all the survivors and updating the Carrier Frequency Offsets (CFOs); and output (520) estimated symbols by tracing back the survivor with least accumulated metric by a length specified as Trace Back Length (TBL).
The method takes the type of modulation of the mixing signals (such as BPSK, QPSK, 8-PSK) and the symbol rate of the mixing signals as inputs. The method works on the assumption that the symbol rates of both the mixing signals are same. Also, the modulation types of both the mixing signals are assumed to be the same. The state table is created at step 502 consisting of all the possible combinations of the symbols (M2 combinations) belonging to the two mixing M-PSK signals. The composite signal (the signal that needs to be separated to two M-PSK signals) is over-sampled and captured at a sampling rate of (4*symbol rate) at data acquisition step 504. At step 506, Root raised cosine (RRC) filtering is applied on the captured data to filter out the noise. At step 508, the received filtered composite data is passed through symbol timing synchronizer which accepts the samples at four times the symbol rate and gives out the time synchronized symbols at the symbol rate using Gardner symbol timing synchronization algorithm.
Joint Maximum Likelihood Sequence Estimation (MLSE) is used for approximating and estimating the transmitted symbol sequences and other unknown parameters such as channel impulse response coefficients and carrier frequency offsets corresponding to the two mixing signals.
Let y_k represent the symbol timing synchronized composite data at the kth instant. y_k can be represented as y_k=?_(l=1)^2¦?_(k_1=-ISI_M)^(ISI_M)¦?s_(l,k+k_1 ) f_(l,k+k_1 ) ? e^(i?_(l,k+k_1 ) )+w_k, where the index l corresponds to the signal l and the index k_1 corresponds to the ISI (Inter Symbol Interference) components in a symbol. s corresponds to the transmitted source symbols.
Post cursor ISI components, pre cursor ISI components (which violate the causality condition) and the current symbol contribution (ignoring the delay between the y_k and s_(l,k)) are represented by negative k_1 values, positive k_1 values and k_1=0 respectively. ISI_M represents the number of post cursor and pre cursor components considered for estimating the maximum likelihood transmitted sequences. The value of ISI_M is decided based on the fading in the channels considered. f represents the channel coefficients, ?_(l,k+k_1 )=2p?f_l (k+k_1 )T corresponds to the carrier frequency offsets (ignoring the initial phases) and w_k represents additive white Gaussian noise (AWGN).
The composite signal separation is achieved by incorporating joint MLSE to accurately estimate the source symbol sequences s_(l,1:N)= [s_(l,1),s_(l,2),….,s_(l,N) ] (the sequences with maximum likelihood) by using the received sequence y_(1:N)= [y_1,y_2,….,y_N ]. While estimating the transmitted symbol sequences, the channel impulse response coefficients and the carrier frequency offsets are also estimated. During the estimation process, N1 survivors are maintained at each symbol (each survivor corresponds to a sequence of transmitted symbols). Each survivor is associated with unique channel impulse response coefficients and unique carrier frequency offsets. The channel coefficients corresponding to the current symbol and the ISI symbols (post cursor and pre cursor) of all the survivors are initialized to be 0.5 and 0 respectively for both the signals. The two carrier frequency offsets of all the survivors are initialized to zero. The accumulated metrics (the accumulated metric corresponds to the likelihood of the survivor) of all the survivors are also initialized to zero. Though the initial estimates are not correct, the estimates will approach the actual values as and when the estimates’ errors are evaluated, and the estimates are updated based on the evaluated errors.
After initialization of all the required parameters, branch metric is computed at step 510. Branch metric from ith survivor to jth state is computed as the sum of the squares of the absolute values of two error components. The first error component corresponds to the difference between the symbol timing synchronized composite data and the post cursor ISI components along with the current symbol components. The second error component corresponds to the difference between the delayed first error component (delayed by ISI_M symbol periods) and the pre cursor ISI components. One of the novelties of this patent includes inclusion of the pre-cursor ISI and post-cursor ISI by the use of two error components. The first and second components are computed as
?error1?_(i,j,k)=y_k-?_(l=1)^2¦?_(k_1=-ISI_M)^0¦?s ^_(l,k+k_1,i) f ^_(l,k+k_1,i) ? e^(i? ^_(l,k+k_1,i) )
and
?error2?_(i,j,k)=?error1?_(i,k-ISI_M)-?_(l=1)^2¦?_(k_1=1)^(ISI_M)¦?s ^_(l,k-ISI_M+k_1,i) f ^_(l,k+k_1,i) ? e^(i? ^_(l,k-ISI_M+k_1,i) )

respectively wherein i, j, k and l represent survivor index, state index, time instant and signal index, respectively. s ^,f ^and ? ^correspond to the estimated symbol sequences, channel impulse response coefficients and carrier frequency offsets, respectively.
Figure 6 illustrates a representation of the branch metric computation from the survivors to the states. As seen in Figure 6, 16 survivors are considered along with QPSK modulation (16 states). The branch metric is calculated as
?_(i,j)=|?error1?_(i,j,k) |^2+|?error2?_(i,j,k) |^2.
At step 512, the accumulated metric (metric which includes summation of all the branch metrics in the surviving path) is computed for each survivor i to enter a state j. It is calculated as
G_(i,j,k)= G_(i,k-1)+?_(i,j).
Out of (N1*N2) possible survivors (N2 = M2, the number of states), N1 survivors are retained using survivors’ selection 514 by sorting the accumulated metrics obtained from step 512 where accumulated metric computation is performed.
Channel impulse response (CIR) coefficients and the carrier frequency offsets are updated for all the selected survivors. The CIR coefficients are updated in step 516 based on the branch error components using least mean squares (LMS) algorithm. The CIR coefficients update rates can be different for the actual symbol component and the ISI components depending on the application. The CIR coefficients are updated using the equation
f ^_(l,(k+1)+k_1,i)=f ^_(l,k+k_1,i)+??error1?_(i,j,k) [s ^_(l,k+k_1,i) ]^* for k_1=0 and negative k_1 values and using the equation
f ^_(l,(k+1)+k_1,i)=f ^_(l,k+k_1,i)+??error2?_(i,j,k) [s ^_(l,k-ISI_M+k_1,i) ]^* for positive k_1 values where * denotes conjugation.
In one embodiment, updating Channel Impulse Response (CIR) coefficients for all the survivors based on the branch error incurred comprises: using the first error component for updating the CIR coefficients corresponding to the current symbol and post cursor ISI components using least mean squares algorithm; and using the second error component for updating the CIR coefficients corresponding to the pre cursor ISI components using least mean squares algorithm.
At step 518, CFO is updated for all the selected survivors by estimating the phase rotations and plays a crucial role in separating the composite signal. The phase rotations of signal 1 and signal 2 are estimated using the equations
??_(1,k)??[(y_k-y ^_(2,k) e^(j? ^_(2,k) ) ) ?y ^_(1,k)?^* e^(-j? ^_(1,k) ) ] and
??_(2,k)??[(y_k-y ^_(1,k) e^(j? ^_(1,k) ) ) ?y ^_(2,k)?^* e^(-j? ^_(2,k) ) ] respectively
where
y ^_(1,k)=s ^_(1,k) f ^_(1,k)and y ^_(2,k)=s ^_(2,k) f ^_(2,k).
These phase rotations are passed through a phase locked loop (PLL) to estimate the phase accurately at the kth instant and the estimated phase information is used for updating the two CFOs corresponding to the two mixing M-PSK signals. The computation of CFOs is not limited to inclusion of the current symbol components alone as the equations depict. Rather, the method involves inclusion of post cursor ISI components for the computation of CFOs depending on the application.
The survivor with least accumulated metric is selected in order to output the separated symbols (with maximum likelihood). The accumulated metric relates to the likelihood of a survivor (the higher the metric, the lower is the likelihood and vice-versa). At step 520, the selected survivor is traced back by length defined as TBL (Trace Back Length) to output the estimated symbols (or the state).
The method discloses multiple ways for implementing trace back. One implementation includes outputting one symbol corresponding to each of the separated signals for every incoming timing synchronized composite symbol. Another implementation includes block processing which involves outputting a block of symbols corresponding to each of the separated signals for a block of symbol timing synchronized data wherein the block length is defined by TBL. Selection of the trace back implementation method depends on the application and the computation time available for real time separation (as the former one needs more computation time than the latter).
In one embodiment, the method discloses that the separated signals output can be hard (wherein the symbols information is output without any noise) or soft (allowing the noise to be passed on to the next stages). One advantage of using soft output includes improvement in the FEC decoder’s error correction capability in most of the cases.
The steps 510-520 are performed for every incoming symbol timing synchronized composite data. The unknowns in the composite signal separation include the maximum likelihood transmitted sequences, the channel impulse response coefficients and the carrier frequency offsets. Among these, carrier frequency offsets are fixed and can be fixed after initial estimation. Once the carrier frequency offsets are estimated, the CFO values are directly incorporated during the signal separation by by-passing the phase rotation estimation and updation step 518.
The other advantage that comes with using pre-estimated CFOs is the reduction in the number of survivors as the number of unknown parameters decreases and also permits real time composite signal separation by by-passing the computationally intensive CFO estimation. Therefore, more number of survivors are maintained at each symbol during the CFO estimation process and less number of survivors are maintained at each symbol for realizing real-time separation (as compared to the number of survivors maintained during the CFO estimation process).
During the real time composite signal separation, the survivors’ selection step 514 utilizes two levels of sorting for reduction in execution time to support real time composite signal separation without affecting the separation capability, wherein the first level of sorting corresponds to the selection of states by sorting the minimum accumulated metrics of the states (minimum accumulated metric of a state is computed as the least of the accumulated metrics from all the survivors to a state); and the second level corresponds to the selection of survivors from the selected states by sorting the accumulated metrics.
During real-time implementation, the survivors having exactly the same set of maximum likelihood transmitted source symbol sequences are redundant, and the redundant survivor is not considered for analysis in the next symbol. Also, it has to be ensured that the accumulated metrics do not overflow when they exceed the maximum permitted values of the variable types. Therefore, the accumulated metrics of all the survivors are subtracted by the least of the accumulated metrics once in every N3 symbols.
The performance of the composite signal separation and the number of bit errors after separation depend on various parameters comprising of the power level differences between the two mixing signals in the composite signal, carrier frequency offset differences between the two mixing signals in the composite signal and the tuning parameters such as the number of survivors maintained at each symbol, ISI_M, trace back length, channel impulse response update rate and carrier frequency offset update rate.
The foregoing description of the invention has been set merely to illustrate the invention and is not intended to be limiting. Since modifications of the disclosed embodiments incorporating the substance of the invention may occur to person skilled in the art, the invention should be construed to include everything within the scope of the invention.

,CLAIMS:
1. A system for single channel blind source separation of M-Phase Shift Keying (M-PSK) signals in Very Small Aperture Terminal (VSAT) interception, the system comprising:
an antenna sub-system (104) configured to receive the signals from the satellite transponder (102);
a front end RF system (106) having a Low Noise Amplifier (LNA) (108) configured to amplify the received signals, a down converter (110) configured to down convert the signal and an Intermediate-frequency (IF) Amplifier (118) configured to reduce noise and intercept the signal efficiently;
a separator hardware (120) configured to receive the composite signal, wherein the composite signal comprises, two M-PSK signals mixed in the same frequency band, and the separator hardware is further configured to output two separated M-PSK signals occupying different frequency bands;
a Radio Frequency (RF) transceiver (124) comprising an Analog to Digital Converter (ADC) (126) configured to capture the composite data;
a Field Programmable Gate Array (FPGA) board (128) comprising a pre-processing unit (130) configured to filter the captured data using root raised-cosine filter and to perform symbol timing synchronization;
a processor based hardware (132) comprising a Carrier Frequency Offset (CFO) estimation unit (134) configured to estimate two carrier frequency offsets corresponding to the two M-PSK signals from the symbol synchronized composite data;
the processor based hardware (132) comprising a signal separation unit (136) configured to estimate the transmitted symbol sequences corresponding to the two M-PSK signals by incorporating the pre-estimated CFO information and by estimating the channel impulse responses of the two channels involved using the symbol synchronized composite data;
the FPGA board (128) comprising a post-processing unit (138) configured to re-modulate the estimated symbol sequences to M-PSK signals centered at two different frequencies as opposed to the composite signal wherein the M-PSK signals are centered at the same frequency; and
the RF transceiver (124) comprising a Digital to Analog Converter (DAC) (140) configured to convert the digital M-PSK waveforms to analog M-PSK signals.

2. The system as claimed in claim 1, wherein the separator hardware (120), comprises:
the Carrier Frequency Offset (CFO) estimation unit 134 configured to estimate the carrier frequency offsets of the two mixing signals before the start of real time signal separation, by maintaining a larger set of survivors at each symbol; and,
the signal separation unit 136 configured to perform real time composite signal separation by incorporating the estimated CFOs from the CFO estimation unit (134) and by maintaining a smaller set of survivors at each symbol.

3. The system as claimed in claim 1, wherein the CFO estimation unit (134) comprises of:
a state table creation unit (302) configured to generate all the possible combinations of the symbols belonging to the two mixing M-PSK signals;
a branch metric computation unit (304) configured to compute the branch metric incurred for a survivor to enter a state;
an accumulated metric computation unit (306) configured to compute the accumulated metric incurred for a survivor to enter a state, wherein the accumulated metric includes summation of all the branch metrics in the surviving path;
a survivors’ selection unit (308) configured to select the best survivors from all the possible survivors;
a Channel Impulse Response (CIR) updation unit (310) configured to update the channel impulse response coefficients for all the survivors; and
a phase rotation estimation and updation unit (312) to update the CFOs by estimating the phase rotations for all the survivors.

4. The system as claimed in claim 1, wherein the signal separation unit (136) comprises:
a state table creation unit (402) configured to generate all the possible combinations of the symbols belonging to the two mixing M-PSK signals;
a branch metric computation unit (404) configured to compute the branch metric incurred for a survivor to enter a state;
an accumulated metric computation unit (406) configured to compute the accumulated metric incurred for a survivor to enter a state, wherein the accumulated metric includes summation of all the branch metrics in the surviving path;
a survivors’ selection unit (408) configured to select the best survivors from all the possible survivors;
a Channel Impulse Response (CIR) updation unit (410) configured to update the channel impulse response coefficients for all the survivors; and
a trace back unit (412) configured to output the estimated maximum likelihood source symbol sequences or state sequences.

5. A method for single channel blind source separation of M-Phase Shift Keying (M-PSK) signals in Very Small Aperture Terminal (VSAT) interception, the method comprising:
creating a state table (502) comprising all the possible combinations of the symbols belonging to the two mixing M-PSK signals;
receiving (504) a composite signal, at a data acquisition unit (504), to process the composite signal;
filtering (506) the received composite data, using a root raised cosine (RRC) filter, to reduce noise;
processing (508) the received filtered composite data and output time synchronized symbols at symbol rate;
computing (510) branch metric incurred for each survivor to enter a state at each time synchronized symbol;
computing (512) accumulated metric incurred for each survivor to enter a state at each time synchronized symbol;
selecting (514) one or more survivors by sorting the accumulated metrics;
updating (516) Channel Impulse Response (CIR) coefficients for all the survivors based on the branch error incurred;
estimating (518) phase rotation and updating phase rotation for all the survivors and updating the Carrier Frequency Offsets (CFOs); and
outputting (520) estimated symbols by tracing back the survivor with least accumulated metric by a length specified as Trace Back Length (TBL).

6. The method as claimed in claim 5, wherein computing branch metric, comprises:
including the pre-cursor Inter Symbol Interference (ISI) and post-cursor Inter Symbol Interference (ISI);
computing branch metric from each survivor to a specific state as the sum of the squares of absolute values of a first error component and a second error component, wherein:
the first error component corresponds to the difference between the symbol timing synchronized composite data and the post cursor ISI components along with the current symbol components;
the second error component corresponds to the difference between the delayed first error component and the pre cursor ISI components.

7. The method as claimed in claim 5, wherein selecting one or more survivors, comprises:
using two levels of sorting for reduction in execution time to support real time composite signal separation without affecting the separation capability, comprising:
computation of minimum accumulated metric of a state as the least of the accumulated metrics from all the survivors to a state;
the first level of sorting for the selection of states by sorting the minimum accumulated metrics of the states; and
the second level of sorting for the selection of survivors from the selected states by sorting the accumulated metrics.

8. The method as claimed in claim 5, wherein updating Channel Impulse Response (CIR) coefficients for all the survivors based on the branch error incurred, comprises:
using the first error component for updating the CIR coefficients corresponding to the current symbol and post cursor ISI components using least mean squares algorithm; and
using the second error component for updating the CIR coefficients corresponding to the pre cursor ISI components using least mean squares algorithm.

9. The method as claimed in claim 5, wherein estimating phase rotation and updating phase rotation for the survivors and updating the Carrier Frequency Offsets (CFOs), comprises:
estimating two carrier frequency offsets corresponding to the two M-PSK signals by estimating the phase rotations and passing the estimated phase rotations through PLL (Phase Locked Loop).