Enhancement of Advanced Range Telemetry (ARTM) Channels via Blind Equalization

Abstract

The Joint Services Advanced Range Telemetry (ARTM) Program at Edwards Air Force Base has been evaluating FQPSK-B for possible upgrades to the existing telemetry equipment. It has been found in the wideband channel sounding experiments sponsored by ARTM that the in-flight fading channel can be modeled as a 3-ray multipath channel[1]. Delay spread for a typical in-flight channel is in the order of 300 nanoseconds. Furthermore, the pre-flight channel is characterized by much more severe multipath, in which the delay spread is in the order of microseconds covering one or more symbols when the FQPSK-B transceiver operates at a rate of millions of symbols per second. This adverse channel condition inevitably causes tremendous distortion in the received signals due to severe inter-symbol interference (ISI) from the multipath. This paper provides an assessment of the potential ability of blind equalization to reduce the FQPSK-B system susceptibility to degradation caused by dynamic frequency selective fading in the aeronautical telemetry environment. In particular, a blind equalizer applique that can be inserted prior to the demodulator without knowledge of the received signal such as carrier frequency, symbol timing and sequence, etc, is proposed. Since it is desired that the equalizer applique operate independently of the carrier frequency and given that the modulation of interest is constant envelope (PCM-FM or FQPSKB), we have selected the constant modulus algorithm (CMA)[2] cost function for implementation. Extensive tests on both simulated and recorded FQPSK-B data transmitted over different ARTM channels have been conducted and the blind equalizer structure has shown substantial improvements, even on the difficult ARTM pre-flight channels. The CMA adapts the equalizer coefficients to minimize the deviation of the output envelope from an arbitrary constant level. This paper depicts the pre-flight and in-flight channel conditions using time and spectral domain measurement. It quantifies the benefit of the blind CMA tapped delay line equalizer. Due to the extensive signal processing requirements associated with the very high sampling rate (100 MHz) of the FQPSK-B system, hardware implementation complexity is very high. Complexity reduction issues regarding the implementation of the CMA using Field Programmable Gate Array (FPGA) will also be presented.