Hybrid Optimization Method for Social Internet of Things Service Provision Based on Community Detection

Abstract

The Internet of things (IoT) and social networks integrate into a new area called the social Internet of things (SIoT). The SIoT is characterized as a social network of things that has enhanced intelligence and social awareness. Essential criteria for both IoT and SIoT networks involve effective service provisioning and the determination of device methods. The discovery of services and selecting the optimal solution to composite them are service provisioning challenges of the SIoT environment. Addressing these challenges requires efficient optimization methods. Traditional optimization algorithms have strengths and weaknesses. For example, a genetic algorithm (GA) can find global optima but suffer from diversity disappearing prematurely, whereas a backtracking search algorithm (BSA) offers better global exploration but converges more slowly. This article proposes a new hybrid optimization algorithm called the improved genetic backtracking search algorithm based on community detection (IGBSA-CD) to overview these limitations. This approach improves the GA's ability and integrates with the advantages of BSA to identify suitable devices to fulfill user requirements by applying the optimized service provision (discovery, selection, and composition) in detected communities. It is based on a new community detection algorithm to reduce the space for service discovery. The experimental results show that the suggested community detection algorithm surpasses current clustering techniques in execution time and cluster quality. IGBSA-CD more rapidly produces solutions that are near-optimal with average success rates of over 96.3% for different sample sizes. The fitness values for each sample size and task also exhibit similar convergence, which stabilizes at 0.2–0.3 after multiple generations. The average response time of IGBSA-CD presents that it is efficient in all three tasks is 0.04 s. It also has a consistently lower response time, even when the task is complex. Furthermore, IGBSA-CD outperforms other optimization approaches in response time and offers superior quality and adaptability within different tasks and sample sizes.