Simulation and Formal Verification in WSNs Empowered by Machine Learning: A Review

The performance analysis of the protocols that were designed in WSNs can be conducted either using simulation or through formal verification. The ability of simulations and formal methods to specify and verify WSNs has attracted the attention of many researchers. Since these methods are practically based on state space exploration, they face some scalability issues when the defined system has a high density, which is frequently the case for WSNs. Machine learning (ML) techniques can be used to address this issue. ML is the technique that automatically learns from data and acts without re-program or human intervention. Applying ML in WSNs to improve the performance of these networks consists of using the historical data obtained by simulation or formal verification with the aim of predicting the future state of the deployed network to limit human intervention. This work reviews recent research works that applied ML techniques in many domains, and after that, works that exploited ML-based algorithms in WSNs are explored. Finally, we present a discussion on open issues.