eduzhai > Applied Sciences > Engineering >

Heterogeneous Task Offloading and Resource Allocations via Deep Recurrent Reinforcement Learning in Partial Observable Multi-Fog Networks

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 16 pages

Abstract: As wireless services and applications become more sophisticated and requirefaster and higher-capacity networks, there is a need for an efficientmanagement of the execution of increasingly complex tasks based on therequirements of each application. In this regard, fog computing enables theintegration of virtualized servers into networks and brings cloud servicescloser to end devices. In contrast to the cloud server, the computing capacityof fog nodes is limited and thus a single fog node might not be capable ofcomputing-intensive tasks. In this context, task offloading can be particularlyuseful at the fog nodes by selecting the suitable nodes and proper resourcemanagement while guaranteeing the Quality-of-Service (QoS) requirements of theusers. This paper studies the design of a joint task offloading and resourceallocation control for heterogeneous service tasks in multi-fog nodes systems.This problem is formulated as a partially observable stochastic game, in whicheach fog node cooperates to maximize the aggregated local rewards while thenodes only have access to local observations. To deal with partialobservability, we apply a deep recurrent Q-network (DRQN) approach toapproximate the optimal value functions. The solution is then compared to adeep Q-network (DQN) and deep convolutional Q-network (DCQN) approach toevaluate the performance of different neural networks. Moreover, to guaranteethe convergence and accuracy of the neural network, an adjustedexploration-exploitation method is adopted. Provided numerical results showthat the proposed algorithm can achieve a higher average success rate and loweraverage overflow than baseline methods.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×