Research Article
Computation Offloading and Security with Q-Learning
@INPROCEEDINGS{10.1007/978-3-030-44751-9_7, author={Songyang Ge and Beiling Lu and Jie Gong and Xiang Chen}, title={Computation Offloading and Security with Q-Learning}, proceedings={IoT as a Service. 5th EAI International Conference, IoTaaS 2019, Xi’an, China, November 16-17, 2019, Proceedings}, proceedings_a={IOTAAS}, year={2020}, month={6}, keywords={Computation offloading System security Q-learning}, doi={10.1007/978-3-030-44751-9_7} }- Songyang Ge
Beiling Lu
Jie Gong
Xiang Chen
Year: 2020
Computation Offloading and Security with Q-Learning
IOTAAS
Springer
DOI: 10.1007/978-3-030-44751-9_7
Abstract
With the rapid development of the technology and wireless communication, the user cannot support the computation-intensive applications, owing to the restricted computation resources, energy supply, limited memory space and communication resources. The emerging computation mode, called mobile edge computing (MEC), provides a solution that the user can unload parts of tasks to edge servers. This communication process should be finished in the wireless network. However, computation offloading in the wireless network can encounter many kinds of attacks. Specifically, edge servers located in the edge of network are vulnerable to these security threats, such as spoofing, jamming and eavesdropping. Moreover, the computation offloading has much time latency and energy consumption. Then, how to minimize this consumption is the another problem to be solved. To improve the security and minimize the consumption, we formulate a system containing a primary user (PU), a second user (SU), an attacker and several edge servers. They communicate with each other by multiple input multiple output (MIMO) technology. In this system, the SU chooses an MEC server from the set of not being occupied by PU, determines an offloading rate and a transmission power, then the attacker selects the action of attack. The aim of this system is to optimize the utility of SU. To solve this problem, a Q-learning based optimal offloading strategy is proposed in dynamic environments. Simulation results show that our proposed scheme can improve the capacity of SU and efficiently decrease the attack rate of the attacker.
