Performance Analysis of Various Techniques with multi objective PSO Based power allocation strategy in cooperative wireless networks

Abstract

newline Designing energy harvesting networks requires modelling of energy distribution under newlinedifferent real-time network conditions. These networks showcase better energy newlineefficiency, but are affected by internal and external faults, which increase energy newlineconsumption of affected nodes. Due to this probability of node failure, and network newlinefailure increases, which reduces QoS (Quality of Service) for the network deployment. newlineTo overcome this issue, various fault tolerance and mitigation models are proposed by newlineresearchers, but these models require large training datasets and real-time samples for newlineefficient operation. This increases computational complexity, storage cost and end-to-end newlineprocessing delay of the network, which reduces its QoS performance under real- time newlineuse cases. To mitigate these issues, this thesis proposes design of a hybrid bioinspired newlinemodel for fault-tolerant energy harvesting networks viafuzzy rule checks. The proposed newlinemodel initially uses a Genetic Algorithm (GA) to cluster nodes depending upon their newlineresidual energy and distance metrics. Clustered nodes are processed via Particle Swarm newlineOptimization (PSO) that assists in deploying a fault-tolerant and energy-harvesting newlineprocess. The PSO model is further augmented via use of a hybrid Ant Colony newlineOptimization (ACO) Model with Teacher Learner Based Optimization (TLBO), which newlineassists in value-based fault prediction and mitigation operations. All bioinspired models newlineare trained-once during initial network deployment, and then evaluated subsequently for newlineeach communication request. After a pre-set number of communications are done, the newlinemodel re-evaluates average QoS performance, and incrementally reconfigures selected newlinesolutions. Due to this incremental tuning, the model is observed to consume lower newlineenergy, and showcases lower complexity when compared with other state-of-the-art newlineii newlinemodels. Upon evaluation it was observed that the proposed model showcases 15.4% newlinelower energy consumption, 8.5% faster communication response, 9.2% better newlinethroughput, and 1.5% better packe