Performance of coherent and noncoherent RAKE receivers with convolutional coding ricean fading and pulse-noise interference

Abstract

The performance of coherent and noncoherent RAKE receivers over a fading channel in the presence of pulse-noise interference and additive white Gaussian noise is analyzed. Coherent RAKE receivers require a pilot tone for coherent demodulation. Using a first order phase-lock-loop to recover a pilot tone with additive white Gaussian noise causes phase distortions at the phase-lock-loop output, which produce an irreducible phase noise error floor for soft decision Viterbi decoding. Both coherent and noncoherent RAKE receivers optimized for additive white Gaussian noise perform poorly when pulse-noise interference is present. When soft decision convolutional coding is considered, the performance degrades as the duty cycle of the pulse-noise interference signal decreases. The reverse is true for hard decision Viterbi decoding, since fewer bits experience interference and bit errors with high noise variance cannot dominate the decision statistics. Soft decision RAKE receiver optimized for pulse-noise interference and additive white Gaussian noise performed the best for both the coherent and noncoherent RAKE receivers. This receiver scales the received signal by the inverse of the variance on a bit-by-bit basis to minimize the effect of pulse-noise interference. The efficacy is demonstrated by analytical results, which reveal that this receiver reduces the probability of bit error down to the irreducible phase noise error floor when pulse-noise interference is present. This demonstrates how important it is to design the receiver for the intended operational environment.