Service caching and user association in cache enabled multi-UAV assisted MEN for latency-sensitive applications

The evolution of 5G (Fifth Generation) and B5G (Beyond 5G) wireless networks and edge IoT (Internet of Things) devices generates an enormous volume of data. The growth of mobile applications, such as augmented reality, virtual reality, network gaming, and self-driving cars, has increased demand for computation-intensive and latency-critical applications. However, these applications require high computation power and low communication latency, which hinders the large-scale adoption of these technologies in IoT devices due to their inherent low computation and low energy capabilities. MEC (mobile edge computing) is a prominent solution that improves the quality of service by offloading the services near the users. Besides, in emergencies where network failure exists due to natural calamities, UAVs (Unmanned Aerial Vehicles) can be positioned to reinstate the networking ability by serving as flying base stations and edge servers for mobile edge networks. This article explores computation service caching in a multi-unmanned aerial vehicle-assisted MEC system. The limited resources at the UAV node induce added problems of assigning the existing restricted edge resources to satisfy the user requests and the associate of users to utilize the finite resources effectively. To address the above mentioned problems, we formulate the service caching and user association problem by placing the diversified latency-critical services to maximize the time utility with the deadline and resource constraints. The problem is formulated as an integer linear programming (ILP) problem for service placement in mobile edge networks. An approximation algorithm based on the rounding technique is designed to solve the formulated ILP problem. Moreover, a genetic algorithm is designed to address the larger instance of the problem. Simulation results indicate that the proposed service placement schemes considerably enhance the cache hit ratio, load on the cloud and time utility performance compared with existing mechanisms. The simulation results show that the proposed content placement improves up to 16.69, and 14.24 per cent on time utility compared with base line algorithms.