QoS-aware hybrid cloudlet placement over joint fiber and wireless backhaul access network

Over the last decades, the emerging paradigms, e.g., centralized mobile cloud computing (MCC), multi-access edge computing (MEC), and collaborative computing offloading (CCO), have attracted extensive attention in adaptive 5G low latency communication for Tactile Internet. Besides, a novel integrated wireless optical transport network emerges in access networks and supports joint fronthaul and backhaul services. Therefore, considering that the task offloading response time, the needs of proximity, and the ultra-dense 5G small cell deployment, a novelty hybrid cloudlet deployment scheme over fiber and wireless backhaul-aware infrastructure is created with efficient cost optimization in mind. In this work, the envisioned Fiber Wireless networks (FiWi) consists of optical backhaul and wireless fronthaul (i.e., integrated access network and backhaul link), whereby base stations (BSs) with fiber and wireless backhauls are referred as wired-BS (WBS) and unwired-BS (UBS), respectively. Therefore, to meet the quality of service (QoS) delay constraints, cloudlets can be deployed in either UBSs, WBS, or remote node (RN). However, we apply mixed-integer line programming (MILP) to resolve the convex optimization problem in terms of minimization deployment cost. Besides, we describe the QoS-aware hybrid cloudlet placement cost algorithm for WBS and UBS coverage areas against different network conditions. It was shown through experimental measurements that the proposed architecture can achieve the scalability in different deployment scenarios. Also, the dependency of minimization cloudlet deployment cost on variable network parameters in terms of user density, network framework, and network QoS can be validated.

Automated summary

This paper introduces a novel hybrid cloudlet deployment scheme that combines fiber and wireless backhaul-aware infrastructure, aiming for efficient cost optimization. Experimental results demonstrate the scalability of the architecture in different deployment scenarios, and the dependency of cost on network parameters.