End-to-end delay is a critical attribute of quality of service (QoS) in cloud computing and computer networks . Service-rate control is a mechanism for providing QoS guarantees in service systems . In thispaper, we introduce a reinforcement learning-based (RL-based) service-rate controller that provides probabilistic upper-bounds . We use queueing theory to model the service systems and use Deep Deterministic PolicyGradient (DDPG) to learn the service rates (action) as a function of the queuelengths in tandem service systems. In contrast to existing RL-basedmethods that quantify their performance by the achieved overall reward, which is hard to interpret or even misleading, our proposed controller provides explicit probabilities on the end-to .end delay of the system . The results are presented for a tandem queueing system with non-exponentialinter-arrival and service times .

Author(s) : Majid Raeis, Ali Tizghadam, Alberto Leon-Garcia

Links : PDF - Abstract

Code :



Keywords : service - learning - systems - tandem - controller -

Leave a Reply

Your email address will not be published. Required fields are marked *