Stochastic Guard-Band-Aware Channel Assignment With Bonding and Aggregation for DSA Networks

Abstract

Fading and shadowing along with the primary user dynamics make channel quality in dynamic spectrum access networks uncertain. Furthermore, the imperfect design of filters and amplifiers in wireless devices motivates the need for guard-bands (GBs) to prevent adjacent-channel interference. In this paper, we develop novel stochastic GB-aware sequential and batch channel assignment schemes that aim at maximizing the spectrum efficiency. In line with recent IEEE 802.11 and LTE standards, our schemes support bonding and aggregation. We propose two assignment models for each of the sequential and batch schemes: a static single-stage and an adaptive two-stage . In the static model, channel assignment is performed once such that the rate demands are probabilistically met. The adaptive model is a two-stage model, where the initial assignment may be corrected once uncertainties are partially revealed. We refer to our formulations of the sequential and batch static assignments as c hance-constrained stochastic s ubset- s um p roblem (CSSP) and c hance-constrained stochastic m ultiple s ubset- s um p roblem (CMSSP) , respectively. Moreover, we develop stochastic formulations for the sequential and batch adaptive assignments, which we refer to as two-stage CSSP with r ecourse (CSSPR) and two-stage CMSSP with r ecourse (CMSSPR) , respectively. Finally, we present computationally efficient simplified versions of CSSP and CSSPR with near-optimal performance.