Search

Online computing offloading is an effective method to enhance the performance of mobile edge computing (MEC). However, existing research ignores the impact of system stability and device priority on system performance during task processing.To address the problem of computing offloading for computing-intensive tasks, an online partial offloading algorithm combining task queue length and energy consumption is proposed without any prior information. Firstly, a queue model of IoT devices is created to describe their workload backlogs and reflect the system stability. Then, using Lyapunov optimization, computing offloading problem is decoupled into two sub-problems by calculating the optimal CPU computing rate and device priority, which can determine the task offloading amount and offloading location to complete resource allocation. Finally, the online partial offloading algorithm based on devices priority is solved by minimizing the value of the drift-plus-penalty function's upper bound to ensure system stability and reduce energy consumption. Theoretical analysis and the outcomes of numerous experiments demonstrate the effectiveness of the proposed algorithm in minimizing system energy consumption while adhering to system constraints, even in dealing with dynamically varying task arrival rates. [ABSTRACT FROM AUTHOR]

System capacity is an important metric for evaluating the performance of edge-caching networks. In most previous studies, the caching content was assumed to remain unchanged in a base station. Hence, the requested probability for each content was treated as stable; In fact, both the number of content and users can change overtime, which impacts the system performance. This article firstly constructs a new user-content evolution model to infer such performance changes, and analyzes the number of links between users and content (i.e., degree). Then, the requested probability of the content and the optimal copy number are calculated by using the content degree. Next, the minimum average transmission distance of the content and system capacity under the cache scheme are given. Finally, a user-content time-varying system is simulated. The results show that the proposed model can characterize the evolution relationship between users and content and achieve better performance in terms of system capacity. [ABSTRACT FROM AUTHOR]

Users can invoke various composite services in Edge Computing (EC) with the development of Service Computing. Generally, users cannot invoke the optimal composite service due to the migration of component service or failure in the edge device. In this study, a Composite Service Provisioning mechanism in EC (CSP-EC) is proposed. It starts when the location of component service changes and terminates when the optimal composite service is obtained. CSP-EC employs dynamic optimization process to meet the real-time requirements. It first caches m optimized intermediate solutions from the first m users, which can be reused by other users to get the final optimal solution. Moreover, a multipopulation Estimation of Distribution Algorithm is used to reduce the probability of falling into a local optimum, and the roulette mechanism is used to accelerate the optimization process. Extensive simulations are conducted based on the real-world dataset. The results show that the proposed mechanism achieves higher quality, better stability, and shorter execution time compared with other schemes. [ABSTRACT FROM AUTHOR]