A novel scheduling framework for heterogeneous traffic in the smart grid

For a resource constrained network, an efficient packet scheduling technique ensures the maximum possible resource utilization while maintaining acceptable network performance at a given time. The scheduling process is closely associated with quality of service (QoS), which provides the customers a certain level of quality assurance while receiving services from the network. For a packet scheduler, the scheduling task becomes even more challenging in a multi-service environment,where the network has to allocate its limited available resources to multiple contending devices/applications with diverse QoS requirements. Hence, optimised distribution of network resources — while still meeting the QoS demands for each traffic class — is the key feature of an efficient packet scheduler.In recent times, the rise of machine-to-machine (M2M) communications, such as the Smart Grids, have further increased the necessity and importance of QoS management and enforcement due to their mission-critical nature. The co-existence of mission-critical M2M traffic and conventional multimedia traffic requires perpacket QoS treatment due to their diverse delay and priority requirements. Consequently,the prevalent practice in conventional networks of supporting session based QoS management is inadequate and a new QoS paradigm is required. To address the aforementioned challenges, this thesis proposes a novel scheduling framework for dynamic resource allocation, where the allocated resource is a function of dynamic queue-sizes of heterogeneous traffic classes and their projected delays. The delay is managed in a manner that the delay-sensitive traffic will be given priority over the delay-tolerant traffic, while still meeting the delay bound and maximum queue size. Thus, it is able to efficiently handle both base and peak traffic loads and provides guaranteed QoS in a multi-service communications network. The performance of the proposed scheduling framework is first validated with Monte-Carlo simulation model based on MATLAB under a generic packet networking environment and compared with two conventional scheduling algorithms. The corresponding results show that the proposed scheduling algorithm achieves significantly better performance compared to the other two schemes. The simulation model is then extended to emulate a multi-service Smart Grid environment to demonstrate the performance in a real-life scenario. To validate the proof-of-concept (PoC), an IEEE 802.16-based WiMAX network was used, which serves both M2M traffic and non-M2M traffic simultaneously. The corresponding results are consistent with those obtained with the generic networking model and demonstrates the applicability of the proposed scheduling framework in a challenging environment, such as the Smart Grid.