Low-power D-BCT decoder for Wi-Max using neural networks

Abstract

The convolution turbo code (CTC) has large memory power consumption. To reduce the power consumption of the state metrics cache (SMC), low-power memory-reduced trace back maximum a posteriori algorithm (MAP) decoding is proposed. Instead of storing all state metrics, the trace back MAP decoding reduces the size of the SMC by accessing difference metrics. The proposed trace back computation requires no complicated reversion checker, path selection, and reversion flag cache. MAP decoders are necessary components of powerful iterative decoding systems such as Turbo codes. For double-binary (DB) MAP decoding, radix-2*2 trace back structures are introduced to provide a trade off power consumption. Trace back MAP decoding: In this, the trace back MAP decoding is proposed to trace the state metrics back by accessing the difference metrics. Trace back convolutional decoding: In the conventional path, the state metrics computed by the natural recursion processor (NRP) in the natural order are stored in the SMC. These two trace back structure, MAP decoding structure achieve power reduction of the state metrics cache. Artificial intelligence technique called Neural network algorithm which is proposed to reduce the power consumption of D-BCT decoder and achieves a better performance in terms of power, latency, etc., as compared to existing methods.