Traditional power grids are currently being transformed into Smart Grids (SGs). SGs feature multi-way communication among energy generation, transmission, distribution, and usage facilities. The reliable, efficient, and intelligent management of complex power systems requires integration of highspeed, reliable, and secure data information and communication technology into the SGs to monitor and regulate power generation and usage. Despite several challenges, such as trade-offs between wireless coverage and capacity as well as limited spectral resources in SGs, wireless communication is a promising SG communications technology. Cognitive Radio Networks (CRNs) in particular are highly promising for providing timely SG wireless communications by utilizing all available spectrum resources. We provide in this paper a comprehensive survey on the CRN communication paradigm in SGs, including the system architecture, communication network compositions, applications, and CR-based communication technologies. We highlight potential applications of CR-based SG systems. We survey CR-based spectrum sensing approaches with their major classifications. We also provide a survey on CR-based routing and MAC protocols, and describe interference mitigation schemes. We furthermore present open issues and research challenges faced by CR-based SG networks along with future directions.Index Terms-Advanced metering infrastructure (AMI), Communication architecture, Cognitive radio network (CRN), Smart grid (SG).1553-877X (c) 2 to license holders on a long-term basis for large geographical regions. This fixed spectrum allocation can result in inefficient spectrum utilization. Cognitive radios (CRs) operating on the principle of Dynamic Spectrum Access (DSA) may be a suitable option to solve these spectrum inefficiency problems [18]. CRs use the existing spectrum through opportunistic access to the licensed bands without interfering with the licensed users. CRs determine the spectrum portions unoccupied by the licensed users known as spectrum holes or white spaces, and allocate the best available channels for CR communication [19], [20].After identification of available spectrum holes, the spectrum utilization can be improved through joint spatial and temporal spectrum sharing. SG applications, especially demand response management, can benefit from joint spatial and temporal spectrum sharing [21]. Machine-to-machine (M2M) communication is an emerging paradigm for providing ubiquitous connectivity between devices to communicate autonomously without human intervention. However, a high number of connected devices can heavily congest the spectrum in existing communication networks. CR technology may be a suitable option for mitigating the potential spectrum shortage arising from M2M communications in the SG [22].
B. Contribution of this Survey ArticleWhile a few overview articles have outline some selected aspects of CR communication in the SG [23]- [25], to the best of our knowledge, there is no prior survey that comprehensively covers CR-based...