1. SCOPE

This document describes the functional requirement of the Bearing Computation and Estimation using FIR filter algorithm's for Tactical Air Navigation System on fighter aircraft's together with Physical specifications.

2. Description

The TACAN is an onboard navigational equipment, which is used to guide the pilot by providing the information about bearing and distance with respect to the ground station. It is a short-range, rho-theta, and ultra high frequency, electronic air navigation system that provides a continuous indication of bearing and distance.The system consists of a pulse transmitter and receiver system (interrogator) carried in the aircraft, which communicates with a pulse receiver-transmitter system (Transponder) positioned on the ground or on a mobile station.

Operational Frequency:

The TACAN system operates in UHF range, L-Band (962MHz - 1213MHz) with 1MHz channel spacing.

Transmitter frequency range - 1025MHz to 1150MHz

Receiver frequency range - 962MHz to 1213MHz

Operational Modes:

The system shall operate in the following modes, T/R mode: In this mode system shall compute distance, bearing and time to go with respect to the ground station.

A/A mode: In this mode system shall compute distance with respect to another aircraft.

REC mode: In this mode, system shall calculate only bearing with respect to the ground station.

The TACAN receiver system computes the bearing of the aircraft by analyzing the
amplitude modulation of the received pulses and by extracting the main and auxiliary
reference pulses transmitted by a ground beacon. The amplitude modulation of the received pulses is caused by the rotation of the beacon's antenna radiation pattern. The received signal has 15Hz and 135Hz modulation on its amplitude as shown in the below.
Figure 1 - TACAN modulation envelope

The beacon sends a pulse pattern (characterized by the interval between pulses and number of pulses) named "Main Reference Burst (MRB)" once every 360 degrees of antenna rotation and sends another pulse pattern named "Auxiliary Reference Burst (ARB)" once every 40 degrees of antenna rotation as shown in below Figure 2 - Cardioid radiation pattern of 15Hz and 13 5Hz signals

The beacon. The beacon also sends "filler" pulse pairs in between these reference at more or less regular interval (of about 740 usee) to facilitate bearing estimation in the receiver as shown in below figure.

The bearing measurement processor computes the coarse bearing estimate from the time interval between the zero crossing of 15Hz modulation component and MRB reference ;

It computes the (more accurate) fine estimate from the interval between the zero crossing of the 135 Hz modulation component and ARB reference. The MRB and ARB patterns are different for the X and Y channels of reception. Each element of a group of pulses may be a twin pulse (two pulses separated by a specific time interval) or a signal pulse as per TACAN specifications.

The beacon also transmits an "Identity" pattern once every 30 seconds to identify itself to the receiver, the receiver may also receive signal from DME beacon in which case the pulses received have constant amplitude and there is no amplitude modulation of the received signal. The processor must compute modulation index to decide whether the received signal is from a DME beacon.

The bearing estimate processor must also analyses the received signal to control the AGC (Automatic Gain Control) function of the receiver to maintain the signal at the Analog to Digital Converter (ADC) between specific levels for the specified RF input dynamic range of the receiver.

The processor must also indicate an "bearing data invalid" signal under specified
conditions.

3. Filter Design Innovation:

The 15 Hz and 135 Hz filters have designed carefully into account extremely narrow bandwidth, high ADC data rate and stringent stop band attenuation requirement and the limited computational power of DSP. The filters will run on sub sampled data , by keeping all these factors in mind . As well as the loss of time resolution due to sub sampling is achieved by a simple process of holding the last known sample value as filter input until a new sample becomes available.

The actual choice of filter is obtained by design space exploration in a Matlab environment keeping the following constraints in mind.

• 15 Hz to 135 Hz filters can be realized within 10MIPS of Computational power.

• Filters are implimented with FIR type so that linear phase relationship (Phase Vs Frequency) can be obtained. The linear phase relationship censures that the filter delay remains constant even if the input modulation frequencies are slightly different from the center frequency

• The filter sampling rate must be in multiples of 6uses.

• The 15Hz filter must reject the DC Component and the 135 Hz component by about 50dB. Similarly the 135 Hz filter must reject the DC component and 15 Hz component by about 50 dB. The level of rejection ensures that the rejected components are buried in noise.

It is enough to run the 15Hz and 135 Hz filters at the rate of 192 and 96 usee respectively.

Time accuracy of 192 usee in 15 Hz filter corresponds to an angle accuracy Of (192 /66667) *360 = 1.03 degree i.e. +/- 0.515 degrees (66667 uses being the period of 15 Hz Cycle)and a time accuracy of 96 usee in 135 Hz corresponds to angle accuracy of (96/7407.4)*40 = 0.52 degrees (i.e. +/- 0.26 degrees). These errors compare favorably with the design goals.We must also keep in mind that the pulses themselves are available on an average around 740 usee (except at the MRB / ARB / ID reference bursts where they are spaced closer). Design space exploration in Matlab environment with these data rates indicates that a 256 tap FIR filter for the 135Hz filter (run every 96usecs) and a 1024 tap FIR filter for 15 Hz filter (run every 192 usee) meet the filter specification. Both of them use a Blackman window. Fig.1 gives the response of the 15Hz filter

The filters are time critical and are implemented in assembly. The filters use single cycle 6multi function instruction to multiply an element of input array with an element of co-efficient. Adding the product to previous sum and simultaneously fetching into relevant registers and coefficient elements needed for the next cycle product multiplication This allows us to use (without considering overheads) one instruction per filter tap. The 15 Hz filters needs therefore 1024 cycles and 135 Hz filter needs 256 cycles per call (Without considering overheads) The overheads will be about 60 cycles . Since the 15 Hz filter is called every 192 usesc the computational load comes to (1024 +60)/192+ 5.65 MIPS approximately. For 135 Hz filter that is called every 96 usee , the computational load is 256+60) 96+3.3 MIPS approximately . As seen in the filters which were implemented in about 9MIPS.

4. FUNCTIONAL SPECIFICATION

CLAIM:

I claim this algorithm

(1) Provides optimum delay to compute and estimation of bearing , it's arround 10 MIPS (500/Jsec). All the signal processing are carried out at 6 usees rate so the extraction of twin pulse / single pulse positions can be realized in 200 usee. All other processing like extraction of reference patterns, computation of coarse and auxiliary bearing and validation checks can be completed with in 200 usee.

(2) Can be implemented with minimum resources on single DSP processor with out
external buffers.

(3) Bearing with ±1° accuracy for RF input level between -82dBm to -20dBm when the overall modulation index 55% max when each 15 Hz & 135Hz modulation index between 12% to 30% for 95% probability.

(4) Can predict the bearing even if the reference signal is Loss for more than 15 sec.