End-to-End High Speed Forward Error Correction Using Graphics Processing Units

摘要

Forward error correction (FEC) is an efficient error recovery mechanism for wireless networks in which erroneous packet is corrected in the destination node. More importantly, real-time and high-speed wireless networks require fast error recovery to ensure quality of service (QoS). Since graphics processing units (GPUs) offer massively parallel computing platform, we propose a GPU-based parallel error control mechanism using extended Hamming code supporting single-bit as well as multiple-bit error correction. We compare the performance of the proposed GPU-based approach with the equivalent sequential algorithm that runs on the traditional CPU for error strength, t, such that 1 ≤t≤7. Experimental results demonstrate that the proposed GPU-based approach outperforms the sequential approach in terms of execution time. Moreover, the proposed parallel implementation yields significant reduction in computational complexity from O(n~3) of the sequential algorithm to O(n) of the GPU-based approach, leading to tremendous speedup gain.