A Context-aware Radio Resource Management in Heterogeneous Virtual RANs

New-generation wireless networks are designed to support a wide range of services with diverse key performance indicators (KPIs) requirements. A fundamental component such networks, and pivotal factor the fulfillment target KPIs, is virtual radio access network (vRAN), which allows high flexibility on control link. However, fully exploit potentiality vRANs, an efficient mapping rapidly varying context decisions not only essential, but also challenging owing interdependence user traffic demand, channel conditions, resource allocation. Here, we propose CAREM, reinforcement learning framework for dynamic allocation in heterogeneous selects best available link transmission parameters packet transfer, so as meet KPI To show its effectiveness, develop testbed proof-of-concept. Experimental results demonstrate that CAREM enables under different settings demand. Also, compared closest existing scheme based neural standard LTE, exhibits improvement one order magnitude loss latency, while it provides 65% latency relatively contextual bandit approach.