Type-based multiple-access with bandwidth extension for the decentralized estimation in wireless sensor networks

Type-based multiple-access (TBMA) is a bandwidth efficient transmission scheme for decentralized estimation in wireless sensor networks (WSNs) over non-orthogonal multiple-access channels, whereas its performance degrades severely in zero-mean fading channels. In this paper, we improve its performance in the fading channels by using more resources for transmitting each type, which is named as TBMA with bandwidth extension (TBMA-BE). This transmission scheme needs no feedback of channel state information as existing solutions, thus it is feasible for large scale WSNs. We then develop an approximate maximum likelihood (ML) estimator, and derive the Cramér-Rao lower bound to reveal how the performance and the bandwidth efficiency are related. It is shown from the simulations that the TBMA-BE transmission scheme with the approximate ML estimator performs fairly well in non-orthogonal multiple-access channels with Rayleigh fading, and is even superior to the orthogonal multiple-access protocol at low communication signal to noise ratio (SNR). It also shows that the TBMA-BE is bandwidth efficient for the typical scenarios of WSNs where the communication SNR is low, since the total bandwidth consumption is proportional to the SNR.