ABSTRACT Visual sensor network (VSN) captures and transmits the visual data from various locations in real‐time monitoring for context aware decision making. Visual sensors have characteristics that make them interesting as sources of information for any process. However, sensors are endowed with low‐power cameras for visual monitoring, and it may also impact the outcome that a visual application delivers to the user, such as spatial coverage, lifetime, and dependability. Additionally, due to the restricted resources available to sensor nodes, meeting both coverage and network lifespan criteria may be impossible. In order to address these drawbacks, we developed a hole detection based on grid‐based triangle approach and optimal link state routing algorithm for efficient data transfer in VSN. Initially, the node is deployed and captures the visual information across the environmental area. After node deployment, the network undergoes hole detection using grid‐based triangle approach–starfish optimization algorithm (GT‐SFOA) to identify the uncovered regions. Then, graph learning–based spatial–temporal graph convolutional neural network (GLSTGCN) is used to overlap detection of sensor nodes based on the similarity calculation. Subsequently, the hole recovery mechanism using optimal localizable k ‐coverage (OLKC) algorithm for reroute similar sensing nodes to cover the holes. Next, low latency optimal link‐state routing (LL‐OLSR) algorithm is used to transmit the sensor data to the base station effectively. The proposed approach achieves a hole discovery time of 3.3 ms for 100 nodes, an average coverage degree of 81.8%, a packet loss of 2% for 100 nodes, and a routing overhead of 96.49%. The proposed approach improves the coverage accuracy in dynamic environments to enhance the reliability of the visual sensor network.
Rajkumar et al. (Fri,) studied this question.