Search

Mobile edge computing (MEC)-enabled heterogeneous networks which compose of multiple layers with different MEC servers can support massive computation-intensive and delay-sensitive applications by pushing cloud functions to the edge of the networks. The heterogeneous nature of networks and the different transmit powers, computing capabilities, coverage areas and various sizes tasks of MEC servers make the resources (such as communication and computation) allocation become more challenging. More importantly, facing the ever-increasing greenhouse gas emission concerns and the rapid growth of the operational cost, how to effectively utilize the resources to improve computation efficiency is of vital importance. In this paper, with the aim to maximize the computation efficiency of the MEC-enabled heterogeneous networks through proper resource allocation, a joint task offloading and resource management problem is formulated. The formulated problem turns out to be a mixed integer nonlinear programming (MINLP) problem. To solve it, an Advantage Actor-Critic (A2C)-based on joint task offloading and resource allocation algorithm (A2C-JTRA) is proposed to determine the offloading strategy, transmit power control and the central processing unit (CPU) computing frequencies allocation of MEC servers and terminal devices (TDs). Simulation results demonstrate that the superiority of the proposed algorithm in improving the computation efficiency compared to the baseline algorithms.

In this article, we investigate the task offloading strategy in a hybrid edge computing networks, where the tasks from end devices can be either executed locally, offloaded to the edge server or forwarded to other friendly devices for processing. In addition, these tasks in system are also assumed to be generated stochastically and with different priorities. With respect to the model, we consider minimizing the total task delay of the system while ensuring that the high priority tasks been completed precedently. To do so, an optimization problem is formulated to determine the task offloading strategy for each task. A deep reinforcement learning approach is designed to solve the problem, where the double deep Q network (DDQN) is employed as the agent module. Simulation results show that the proposed algorithm achieves 25% higher utility than the greedy one. In addition, the performance is only 11% lower compared to the optimal solution given by exhaustive search, which confirms the effectiveness of the proposed algorithm.

Although edge caching assisted adaptive bitrate (ABR) video streaming has significantly reduced amount of video data to be transmitted via backhaul link, its performance is severely limited by the caching capacity and user heterogeneous preferences. To address this problem, this article investigates a collaborative recommendation, caching and transcoding approach to boost caching performance. A caching hit ratio maximization problem is formulated by designing recommendation, caching and transcoding simultaneously. The problem is an integer non-linear programming (INLP) problem and is NP-hard. To solve it effectively, we decompose it into three subproblems, i.e., content recommendation, caching placement and transcoding decision subproblems. The optimal solution to the first subproblem is derived theoretically, the second is proven to be a monotone submodular maximization problem, which is solved by greedy algorithm, and the third is a knapsack problem and is solved by dynamic programming approach. Finally, a joint recommendation, caching and transcoding (JRCT) algorithm is established to solve these three subproblems iteratively. Extensive simulations are conducted on a real world viewing dataset. It is demonstrated that JRCT can achieve higher caching hit ratio and traffic offloading ratio than the state-of-the-art baselines.

Edge content caching has emerged as a promising solution against network latency by pre-caching popular contents at the edge of networks. However, the edge content caching and wireless resource allocation are intertwined with each other. The content downloading latency depends not only on the edge content caching strategy, but also highly on the wireless resource allocation strategy. To cope with them as a whole, this work aims at jointly optimizing the edge content caching and power allocation to minimize the content downloading latency. In doing so, a joint edge content caching and power allocation problem is formulated. But, this problem is NP-hard and difficult to be solved. Solving it with the traditional approaches will cause quite long computational delay, which cannot meet the real-time requirements of resource scheduling. To address this challenge, a novel deep learning (DL)-enabled joint edge content caching and power allocation framework is proposed. In particular, the formulated problem is firstly transformed into a classification problem in DL field. After that, the convolutional long short-term memory networks (ConvLSTM) are utilized to capture the temporal-spatial features of content requests, and the fully-connected networks (FC) are introduced to explore the users' location differences. Finally, the temporal-spatial features of content requests and the location differences are fused to make intelligent decisions for joint edge content caching and power allocation. Simulation results show that the proposed joint edge content caching and power allocation strategy outperforms the state-of-the-art baselines at about 2.3% in terms of the average content downloading latency while guaranteeing real-time resource scheduling.

Vehicular edge computing (VEC) has been reported as a new computation paradigm to meet the low-latency requirement in vehicular networks. In this article, we study a novel location-aware task offloading mechanism in a VEC-based single-vehicle multi-cell (SVMC) scenario, where the task can be equally partitioned into multiple subtasks. Different from existing work, task uploading and computing are taken into account in a parallel way. Taking the impact of the uncertainty of vehicle location on task uploading time into account, single-cell offloading and multi-cell offloading are investigated, respectively. Hence, the scheduling problem is studied with the objective of minimization task processing delay by jointly designing the amount of offloaded subtasks for multiple cells, where the task offloading decision over the small timescale is investigated due to small-scale fading. The problem turns out to be a min-max optimization problem, which can be transformed into a minimum problem of the absolute value function. For single-cell offloading, a low-complexity multi-time slot offloading (MTSO) algorithm is proposed by jointly optimizing the amount of offloaded subtasks for multiple time slots. For multi-cell offloading, a multi-cell and multi-time slots offloading (MCMTSO) algorithm is proposed by jointly optimizing the amount of offloaded subtasks for multiple time slots in multiple cells with low complexity. Simulation results review that the proposed algorithm can effectively reduce the task processing delay. For single-cell offloading, the task processing delay of MTSO is reduced by 40.5% compared to partial offloading (PO), while for multi-cell case, the MCMTSO scheme can reduce the task processing delay by 24.3% compared to PO.

This article focuses on low-latency transmissions for mobile edge computing (MEC)-assisted adaptive bitrate (ABR) video streaming. With the aim to realize minimum video service latency, an optimization problem is formulated by jointly designing caching, computing and power allocation, subject to diverse video version downloading rate requirements and the limited resources of MEC servers. The formulated optimization problem turns out to be a mixed integer nonlinear programming (MINLP) problem. To solve the problem and satisfy the low-latency service requirement of ABR video streaming, a low-complexity two-step iterative algorithm is proposed by decomposing the MINLP problem into two subproblems, a joint caching and computing subproblem and a power allocation subproblem. Due to the highly coupled nature of the caching and computing, the first subproblem is an integer nonlinear programming (INLP) problem. To solve it efficiently, the INLP problem is equivalently transformed into an integer linear programming (ILP) problem, which is thereafter solved by MATLAB intlinprog function. The second subproblem is a convex problem and is solved by low-complexity power allocation (LCPA) algorithm. Intensive simulation results show that the proposed algorithm converges in a very short time, and can significantly reduce the average latency compared to the state-of-the-art baselines.