WSN nodes location
This is a brief summary of my research on AI-based node localization in wireless sensor networks using a hybrid particle swarm optimization and simulated annealing algorithm.
Introduction
- In the 5G network architecture, IoT will interconnect devices with various potentials during the identical heterogeneous network
- Wireless Sensor Networks (WSN) is the core technology of the IoT.
- Large-scale WSN positioning in the IoT, that is, to locate unknown nodes through the information of known nodes, is a key technology to solve entity position problems in specific environments (such as environmental monitoring and intrusion positioning)
Preliminary
Particle Swarm Optimization
- High efficiency, good robustness and easy convergence.
- But easy to fall into local optimization
Simulated Annealing
- Can jump out of the local optimal solution with a certain probability
Methodology
- Introduce the SA algorithm to break through the local optimal characteristics
- Improve the acceptance process of PSO algorithm in individual extreme value update
Experiments & Discussions
SA-PSO vs Traditional PSO
- Faster convergence
- Better coverage
- More reasonable positioning
Critical Analysis
- Existing Flaws
- PSO traditional inertia weight
- PSO fixer learning Factor
- SA inner loop
- Improvement Directions
- PSO dynamic inertia weight
- PSO contraction learning factor
- SA inner loop with adaptive sampling stability condition
- Large scale node location (number, range) conditions should be considered (already been implemented in my publication)
