Online Resource Management in Energy Harvesting BS Sites through Prediction and Soft-Scaling of Computing Resources

Abstract

Multi-Access Edge Computing (MEC) is a paradigm for handling delay sensitive services that require ultra-low latency at the access network. With it, computing and communications are performed within one Base Station (BS) site, where the computation resources are in the form of Virtual Machines (VMs) (computer emulators) in the MEC server. MEC and Energy Harvesting (EH) BSs, i.e., BSs equipped with EH equipments, are foreseen as a key towards next generation mobile networks. In fact, EH systems are expected to decrease the energy drained from the electricity grid and facilitate the deployment of BSs in remote places, extending network coverage and making energy self-sufficiency possible in remote/rural sites. In this paper, we propose an online optimization algorithm called ENergy Aware and Adaptive Management (ENAAM), for managing remote BS sites through foresighted control policies exploiting (short-term) traffic load and harvested energy forecasts. Our numerical results reveal that ENAAM achieves energy savings with respect to the case where no energy management is applied, ranging from 56% to 66% through the scaling of computing resources, and keeps the server utilization factor between 30% and 96% over time (with an average of 75%). Notable benefits are also found against heuristic energy management techniques.

Adaptive Resource Management for a Virtualized Computing Platform within Edge Computing

“…In [6], computing resources are provisioned depending on the expected server workloads via a reinforcement learning-based resource management algorithm, which learns the optimal policy for dynamic workload offloading and servers autoscaling. Our previous works in [7] and [17], focus on the provision of computing resources (VMs) based on a Limited Lookahead Control (LLC) policy and the network impact (the use of traffic load as a performance metric [18]), after forecasting the future workloads and harvested energy. A single Base Station (BS) optimization case is considered for an off-grid site in [7], and a multiple BS optimization case, each BS site powered by hybrid energy sources, is studied in [17] where the edge management procedures are enabled by an edge controller.…”

“…Our previous works in [7] and [17], focus on the provision of computing resources (VMs) based on a Limited Lookahead Control (LLC) policy and the network impact (the use of traffic load as a performance metric [18]), after forecasting the future workloads and harvested energy. A single Base Station (BS) optimization case is considered for an off-grid site in [7], and a multiple BS optimization case, each BS site powered by hybrid energy sources, is studied in [17] where the edge management procedures are enabled by an edge controller. This work differs from our previous works as the MEC server is placed in proximity to a BS cluster, and not one co-located for each BS.…”

“…Considering that θ CPU (t) is related to the number of VMs running in time slot t, named M (t) ≤ M , and on the CPU frequency that is allotted to each VM, θ CPU (t) is obtained using the linear relationship between the CPU utilization contributed by VM m, and the energy drained is, inspired by [7]:…”

“…Note that this relationship accounts for the amount of energy drained θ MEC (t), that harvested H(t) and that purchased from the electrical grid E(t), which together lead to the next buffer level B(t + 1) through Eq. (7). The online resource management algorithm, ARCES, finds the best control action vector that yields the desired energy savings within the computing environment.…”

“…3.5 in this reference). It is observed in the literature that most of the existing Energy Saving (ES) studies have involved autoscaling [6] [7][8] [9] [10] (scaling up/down the number of computing nodes/servers or VMs), VM migration [11] (movement of a VM from one host to another) and soft resource scaling [12] (shortening the access time to physical resources), all hereby referred to as VM soft-scaling, i.e., the reduction of computing resources per time instance. Hence, the proposed energy models (see [13] for a summary of the proposed models) and the overall operational expenditure of the computing node is usually related to the computation process (i.e., the running VMs), overlooking the communication processes within the server.…”