A cognitive WSN framework for highway safety based on weighted cognitive maps and Q-learning

Wireless technology can provide new efficient techniques for improving highway safety. Wireless Sensor Networks (WSN) have been identified as a key enabling technology for monitoring road conditions and providing early warning messages to drivers about any dangers that may be present. In order to ensure that the end-to-end goals of such a WSN are achieved, this paper proposes a cognitive framework based on the mathematical tools known as Weighted Cognitive Maps (WCM) and Q-learning. WCM is used to design a reasoning machine that can consider multiple conflicting constraints with low complexity. On the other hand, the Q-learning algorithm is used to design a learning protocol that can build a knowledge base which enables the WCM system to make more informed decisions. Thus, a reward system is developed that directly addresses the end-to-end goals of the system. The performance of the cognitive framework is evaluated using extensive computer simulations and compared to state of the art systems. Simulation results show significant performance improvements with the proposed cognitive framework.