Abstract
Conference Title: 2016 IEEE Globecom Workshops (GC Wkshps) Conference Start Date: 2016, Dec. 4 Conference End Date: 2016, Dec. 8 Conference Location: Washington, DC, USA In this paper, we present a joint delay-efficient channel assignment and packet scheduling scheme for a heterogeneous M2M uplink system. The uplink sensor traffic is aggregated at multiple M2M aggregators (MAs) and then processed at a M2M application server (AS). The system bandwidth is divided into multiple orthogonal subcarriers to be assigned to different MA-AS wireless links. We categorize the uplink traffic into multiple classes based on the maximum packet delay, arrival rate and payload size. The delay-requirements of each class are mapped onto sigmoidal utility functions. We then determine the joint subcarrier assignment and packet scheduling scheme such that it maximizes a proportionally-fair system utility metric. The centralized optimization is a mixed integer non-linear programming (MINLP) problem with high computational complexity. Therefore, we propose a two-stage distributed optimization problem that is solved independently across MAs and AS. We first determine the delay-efficient subcarrier assignment and then determine the corresponding delay-efficient packet scheduler. Using Monte-Carlo simulations, we numerically show that distributed optimization does not result in any loss in delay-performance and also outperforms other state-of-the-art schemes.