2019/10/23 by Joan S. Pujol Roig, Roig, Joan S Pujol, David M. Gutierrez-Estevez +3 · 1 citation
Computer Science · Engineering · #Advanced Memory and Neural Computing #FOS: Computer and information sciences #Information Theory (cs.IT) #Networking and Internet Architecture (cs.NI) #Software-Defined Networks and 5G
paper · pdf · doi:10.48550/arxiv.1910.10695
openalex publication_date 2019/10/23 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Management and orchestration (MANO) of resources by virtual network functions\n(VNFs) represents one of the key challenges towards a fully virtualized network\narchitecture as envisaged by 5G standards. Current threshold-based policies\ninefficiently over-provision network resources and under-utilize available\nhardware, incurring high cost for network operators, and consequently, the\nusers. In this work, we present a MANO algorithm for VNFs allowing a central\nunit (CU) to learn to autonomously re-configure resources (processing power and\nstorage), deploy new VNF instances, or offload them to the cloud, depending on\nthe network conditions, available pool of resources, and the VNF requirements,\nwith the goal of minimizing a cost function that takes into account the\neconomical cost as well as latency and the quality-of-service (QoS) experienced\nby the users. First, we formulate the stochastic resource optimization problem\nas a parameterized action Markov decision process (PAMDP). Then, we propose a\nsolution based on deep reinforcement learning (DRL). More precisely, we present\na novel RL approach called, parameterized action twin (PAT) deterministic\npolicy gradient, which leverages an actor-critic architecture to learn to\nprovision resources to the VNFs in an online manner. Finally, we present\nnumerical performance results, and map them to 5G key performance indicators\n(KPIs). To the best of our knowledge, this is the first work that considers DRL\nfor MANO of VNFs' physical resources.\n