Abstract
The service of mobile network develops rapidly nowadays, which generates various computing and resource-intensive applications, such as Internet of vehicles and virtual reality. Mobile edge computing (MEC) is close to data source and users, so terminals can execute tasks at the edge of network. In this way, the heavy load on core network can be relieved and tasks can be executed effectively. However, the demands of users vary from each other and users move all the time. It is difficult for the existing way of service supply to meet demands of all users. Cognitive Agent (CA) is put forward in this paper to help users cache and execute tasks on MEC in advance. In detail, CA is used to build personalized model combined with users’ behavior data. At the same time, it uses Long short-term memory neural network to forecast the moving trajectory of terminal equipment and the service types to be requested, uses the prediction result to generate caching strategy, cache business and shorten the delay of task execution. Besides, to further reduce the stress on MEC, we propose the collaboration of computing, communicating and caching resource with neighboring users’ equipment. To verify the effectiveness of CA, we build a model that assesses the performance of the system. Finally, we design a simulation experiment to execute resource request and resource collaboration. The result of the experiments show that CA can improve the efficiency of communication network, relieve the stress on network and improve the quality of services to users.
•Cognitive agent (CA) is put forward in this article to realize cognition function on mobile edge computing (MEC) in smart devices aiming at fifth generation of the mobile communication system. In detail, CA is used to build personalized model of users’ behavior data.•The prediction model of users’ behavior trajectory and application based on the aware ability of CA, show the cognition of agent. Based on the predicting result, the caching strategy and cache business are generated to shorten the delay of task execution.•To verify the effectiveness of CA, a model is formulated about resource caching and collaboration including hit rate and delay of task execution.•Simulation experiments are performed to prove that CA can well recognize and help terminal equipment execute resource caching and collaboration and the quality of experience and service are improved.