We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:


References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Networking and Internet Architecture

Title: Deep Reinforcement Based Optimization of Function Splitting in Virtualized Radio Access Networks

Abstract: Virtualized Radio Access Network (vRAN) is one of the key enablers of future wireless networks as it brings the agility to the radio access network (RAN) architecture and offers degrees of design freedom. Yet, it also creates a challenging problem on how to design the functional split configuration. In this paper, a deep reinforcement learning approach is proposed to optimize function splitting in vRAN. A learning paradigm is developed that optimizes the location of functions in the RAN. These functions can be placed either at a central/cloud unit (CU) or a distributed unit (DU). This problem is formulated as constrained neural combinatorial reinforcement learning to minimize the total network cost. In this solution, a policy gradient method with Lagrangian relaxation is applied that uses a stacked long short-term memory (LSTM) neural network architecture to approximate the policy. Then, a sampling technique with a temperature hyperparameter is applied for the inference process. The results show that our proposed solution can learn the optimal function split decision and solve the problem with a $0.4\%$ optimality gap. Moreover, our method can reduce the cost by up to $320\%$ compared to a distributed-RAN (D-RAN). We also conclude that altering the traffic load and routing cost does not significantly degrade the optimality performance.
Comments: This paper has been accepted in IEEE International Conference on Communications Workshops (ICC Workshops) 2021
Subjects: Networking and Internet Architecture (cs.NI)
DOI: 10.1109/ICCWorkshops50388.2021.9473703
Cite as: arXiv:2105.14731 [cs.NI]
  (or arXiv:2105.14731v1 [cs.NI] for this version)

Submission history

From: Fahri Wisnu Murti [view email]
[v1] Mon, 31 May 2021 06:51:26 GMT (1001kb,D)

Link back to: arXiv, form interface, contact.