System Power Minimization to Access Non-Contiguous Spectrum in Cognitive Radio Networks

TL;DR

This paper addresses the challenge of minimizing system power in multi-hop cognitive radio networks accessing non-contiguous spectrum by proposing joint optimization algorithms that consider spectrum span and power control.

Contribution

It introduces a novel cross-layer optimization framework with branch and bound and greedy algorithms for power, spectrum span, scheduling, and routing in NC-OFDMA networks.

Findings

System power reduced by 30% using proposed algorithms.

Trade-off characterized between spectrum span and power consumption.

Efficient algorithms outperform classical power minimization methods.

Abstract

Wireless transmission using non-contiguous chunks of spectrum is becoming increasingly important due to a variety of scenarios such as: secondary users avoiding incumbent users in TV white space; anticipated spectrum sharing between commercial and military systems; and spectrum sharing among uncoordinated interferers in unlicensed bands. Multi-Channel Multi-Radio (MCMR) platforms and Non-Contiguous Orthogonal Frequency Division Multiple Access (NC-OFDMA) technology are the two commercially viable transmission choices to access these non-contiguous spectrum chunks. Fixed MC-MRs do not scale with increasing number of non-contiguous spectrum chunks due to their fixed set of supporting radio front ends. NC-OFDMA allows nodes to access these non-contiguous spectrum chunks and put null sub-carriers in the remaining chunks. However, nulling sub-carriers increases the sampling rate (spectrum span) which, in turn, increases the power consumption of radio front ends. Our work characterizes this trade-off from a cross-layer perspective, specifically by showing how the slope of ADC/DAC power consumption versus sampling rate curve influences scheduling decisions in a multi-hop network. Specifically, we provide a branch and bound algorithm based mixed integer linear programming solution that performs joint power control, spectrum span selection, scheduling and routing in order to minimize the system power of multi-hop NC-OFDMA networks. We also provide a low complexity (O(E^2 M^2)) greedy algorithm where M and E denote the number of channels and links respectively. Numerical simulations suggest that our approach reduces system power by 30% over classical transmit power minimization based cross-layer algorithms.