Wireless Sensor Network Hop Count Theory Modeling and Deep Learning Prediction

Wireless Sensor Networks (WSNs) are utilized as distributed systems for wireless communication and data collection from sensors. A WSN can be represented as a graph consisting of sensors and wireless links. If this graph has a path connecting any source sensor to any destination sensor, it is considered a connected network graph. Otherwise, it is a disconnected graph. The connectivity of a network graph is crucial for its proper functioning and applications. Network connectivity can be affected if a certain number of sensor nodes are missing from the topology. This number of nodes can be used as a threshold for connectivity. In a WSN, sensors communicate with their direct neighbors based on the quality of their communication links. This paper proposes a mathematical model to address connectivity and link probability issues in WSNs. The model examines the impact of varying link probability and the number of deployed sensors on network connectivity. The model is evaluated using different radio models in the Cooja emulator with the Rime stack. A multi-hop WSN is implemented as a linear path with various radio models to analyze network connectivity probability and estimate the maximum hop count. Additionally, the success of flood route estimation based on link probability is investigated. Network performance metrics such as packet delivery ratio and packet loss rate are used to evaluate network connectivity and measure the maximum hop count for both radio models. Furthermore, the collected simulation dataset is employed in Deep Learning models to predict the hop count based on link quality parameters.