Erasure Coding for Ultra-Low Power Wireless Networks

Abstract

In this paper, we study erasure coding for ultra-low power wireless networks with power consumption in order of milliwatts. We propose sparse parallel concatenated coding (SPCC) scheme, in which we optimize sparsity and ratio of coded packets over GF(2) (i.e., Galois field of size two) and larger field size such as GF(32) for different values of k so that the total energy cost of the network is minimized. While high sparsity decreases energy cost of encoding, it comes at the tradeoff cost of high reception redundancy. The use of GF(2) packets minimizes the computational cost of encoding and decoding, while the use of small fraction of packets over GF(32) minimizes reception redundancies. Testbed implementation shows that SPCC energy gain increases with increasing packet generation size k. We show that for the case where k}~\leq 40, SPCC reduces energy cost by up to 100% compared with the next best performing coding scheme.