Evaluating positioning estimation using LSE, gradient descent, and particle swarm optimization with dynamic inertia weight in wireless sensor network.

Knowing the position of the interesting object has gotten much attention from many researchers due to the opportunity to develop myriad applications. In this paper, we will integrate all the processes required as a process that can be used to estimate the unknown object's position. First, we measured a real-time signal strength (SS) received by the unknown node simultaneously from three anchor nodes. We use the Tmote sky node as transceivers. Second, we estimate the distance of an unknown node from the anchor nodes based on the path loss model; third, we use the linear least square estimation, gradient descent (GD), and particle swarm optimization (PSO) to estimate four different unknown node position (P1, P2, P3, and P4). We found that all algorithms are sufficient to estimate the node position even with some error in the distance estimation obtained except that linear least square got the worst position estimation. A linear LSE does not minimize the error of unknown node position estimation from all anchor nodes. We found that the PSO algorithm is better than GD algorithms regarding time execution and gives an opportunity to be implemented in real-time hardware execution. [ABSTRACT FROM AUTHOR]