Convolutional Code Performance in Impulsive or Burst Noise Channels

We consider the performance of short constraint-length convolutional codes in conjunction with coherent BPSK modulation in impulsive or burst noise channels when an imperfect erasure mechanism is used to exorcise symbols contaminated by noise hits. The erasure mechanism is a simple holepuncher operating on the sampled matched filter outputs and is characterized by a fixed false alarm probability, PF, and a miss probability, PM. This scheme is representative of a large number of erasure declaration strategies. Some general conclusions are drawn on the basis of cutoff rate arguments under an idealized channel modeling assumption. For short constraint-length convolutional codes we provide tight upper bounds on bit error probability performance under the same idealized channel modeling assumptions. These bounds compare favorably with simulation results for both randomly occurring and periodic noise hits. Departures from idealized channel modeling assumptions are investigated through simulation. Results indicate an important distinction between the case of random and periodic noise hits and has implications for the use of interleaving an impulsive or burst noise channels.