DAOSVM: A Machine Learning Approach to Obstacle-Aware Path Planning in Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are widely used for monitoring and collecting data in areas like agriculture, environmental tracking, and industrial systems. One common issue in these networks is that some sensor nodes end up handling more traffic than others, causing them to run out of energy faster. To solve this, mobile sinks devices that move through the network to collect data have been introduced. These mobile sinks help balance the load, reduce energy use, and improve overall network performance. However, most existing solutions assume that the mobile sink can move freely without any barriers. In reality, physical obstacles such as buildings, walls, or rough terrain can block its path, making data collection more difficult. To overcome this challenge, we introduce a smarter approach called DAOSVM (Data Acquisition with Obstacle-awareness using Support Vector Machine). This method helps the mobile sink make better decisions about where to go and how to get there, even in environments with obstacles. The process has two main steps. First, we use a clustering method called Fuzzy C-Means (FCM) to find the most useful spots for the sink to visit. It looks at where sensor nodes are located and how much battery they have left to choose efficient gathering points. Next, we use the A* pathfinding algorithm to plan the sink’s route. This path avoids obstacles and is guided by a model that considers factors like energy levels, how spread out the nodes are, and how many obstacles are nearby. We tested this method through simulations and found that DAOSVM performs better than existing techniques. It results in shorter travel paths, lower energy use, better data delivery, and greater adaptability in challenging environments. This makes DAOSVM a practical and reliable option for real-world WSN applications where obstacles are a concern.