Delay-Efficient Multiclass Packet Scheduler

Abstract

Over the past decade, machine-to-machine (M2M) communications has rose to prominence as it is increasingly being used for automation of various residential and industrial applications such as vehicle tracking, smart grid, industrial automation, etc. The sheer enormity of M2M uplink traffic has resulted in heavy contention for the limited computational and communication resources, making it even tougher to satisfy the QoS requirement of real-time traffic. This further accentuates the need for QoS-aware packet schedulers at the intermediate traffic aggregators and the application server. However, existing packet schedulers either fail to account for heterogeneity in M2M traffic characteristics and QoS requirements or they are designed for specific wireless technology such as LTE, Zigbee etc. Therefore, in this chapter, we first segregate the heterogeneous M2M traffic into multiple traffic classes and map the delay requirements onto utility functions. We then propose a multiclass delay-efficient packet scheduling heuristic that maximizes a proportionally fair system utility metric by exploiting the knowledge of shape of utility functions. We ensure that the proposed packet schedulers have low implementation complexity, provide proportional fairness among traffic of various QoS classes. Also, the proposed packet scheduler is agnostic to the M2M application, communication standard and hardware-software architecture. Lastly, for validation purposes, we benchmark the fairness and delay performance of proposed packet schedulers against the state-of-the-art packet schedulers.