eduzhai > Applied Sciences > Engineering >

Resource Allocation via Model-Free Deep Learning in Free Space Optical Communications

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 30 pages

Abstract: This paper investigates the general problem of resource allocation formitigating channel fading effects in Free Space Optical (FSO) communications.The resource allocation problem is modeled as the constrained stochasticoptimization framework, which covers a variety of FSO scenarios involving poweradaptation, relay selection and their joint allocation. Under this framework,we propose two algorithms that solve FSO resource allocation problems. We firstpresent the Stochastic Dual Gradient (SDG) algorithm that is shown to solve theproblem exactly by exploiting the strong duality but whose implementationnecessarily requires explicit and accurate system models. As an alternative wepresent the Primal-Dual Deep Learning (PDDL) algorithm based on the SDGalgorithm, which parametrizes the resource allocation policy with Deep NeuralNetworks (DNNs) and optimizes the latter via a primal-dual method. Theparametrized resource allocation problem incurs only a small loss of optimalitydue to the strong representational power of DNNs, and can be moreoverimplemented without knowledge of system models. A wide set of numericalexperiments are performed to corroborate the proposed algorithms in FSOresource allocation problems. We demonstrate their superior performance andcomputational efficiency compared to the baseline methods in both continuouspower allocation and binary relay selection settings.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×