The smart grid relies on communication networks among millions of measuring devices, including remote terminal units (RTUs) and phasor measurement units (PMUs) in grid networks and smart meters on customer premises. Thus, building the communication infrastructure for the smart grid is an expensive and resource-consuming job. Wireless mesh networks have been considered as a cost-effective infrastructure for the smart grid communication network.

This paper investigates whether existing mesh network routing protocols are suitable for delivering smart grid traffic in line with quality of service (QoS) requirements. To drive the QoS requirements, a number of smart grid use cases referenced in a document from the National institute of Standards and Technology (NIST) are classified into advanced metering infrastructure (AMI) functions and supervisory control and data acquisition (SCADA) functions based on network requirements. As a result, requirements for delay and reliability for each function group are presented. For evaluation, the QoS metrics with existing mesh routing protocols are compared through simulation. The authors insist that the ad-hoc on-demand distance vector (AODV) protocol shows better performance than the hybrid wireless mesh network protocol (HWMP) in a large network in terms of throughput, delay, jitter, and packet delivery fraction. When a WiMAX network is used, adaptive modulation coding shows better performance than a single modulation coding scheme.

The major contribution of this paper is its association of the network requirements of smart grid use cases with existing mesh network performance. However, it should be investigated whether the given network requirements indicate network performance in the routing layer. Since the given delay requirement in the paper has a wide range--up to four hours in the case of AMI meter reading--it seems that it reflects various delay situations, such as network disconnection and reattempt in the application layer, that can happen in real systems. The simulation study for the routing layer cannot reflect such delay situations. In the case of the reliability requirement, it must reach up to 99.5 percent, and is not evaluated in the simulation correctly. A similar metric is the packet delivery fraction; however, it does not consider delivery success through retransmission in a higher layer. This is why the simulation shows a poor packet delivery fraction, which does not meet the reliability requirement in most scenarios.

Since all routing protocols and coding schemes in the simulation study show a short delay and a poor packet delivery fraction compared to the given delay and reliability requirements, it is difficult to tell which routing protocol and coding scheme should be used.