Analysis of quantization effects on performance of hierarchical cooperation schemes in Ad-hoc wireless networks

In this paper, we analyze the role of the number of quantization bits on the third phase of the hierarchical cooperation scheme proposed by Ozgur et al. in 2007. The hierarchical cooperation scheme has a digital architecture and requires quantization of MIMO (Multiple-Input Multiple-Output) observations in the third phase. The choice of the number of quantization bits Q has direct effect on the pre-constant factor of the network throughput. By increasing the number of quantization bits, the MIMO capacity in the second phase increases at the expense of higher computational complexity of the third phase. We investigate such trade-off and show that there is an optimum number of quantization bits which maximizes the throughput. In addition, it will be shown that due to quantization noise, it is not possible to increase throughput pre-constant factor as large as desired by increasing transmit power. As a result, there is a saturation value for the pre-constant value. Finally, we will demonstrate that by proper scaling of Q as Q = β log(SNR) for any β independent of SNR, an increasing pre-constant factor of order √(log (SNR)) can be achieved.