MACHINE LEARING FOR MOBILE EDGE COMPUTING

Shangguang Wang

Qiang Duan

Kok-Seng Wong

Claudio A.Ardagna

In recent years,mobile edge computing has attracted a considerable amount of attention from both academia and industry through its many advantages(such as low latency,computation efficiency and privacy) caused by its local model of providing storage and computation resources.In addition,machine learning has become the dominant approach in applications such as industry,healthcare,smart home,and transportation.All of these applications heavily rely on technologies that can be deployed at the network edge.Therefore,it is essential to combine machine learning with mobile edge computing to further promote the proliferation of intelligent edges.In general,machine learning relies on powerful computation and storage resources for superior performance,while mobile edge computing typically provides limited computation resources locally.To this end,the implementations of machine learning algorithms should be revisited for mobile edge computing.

This special issue aims to become a valuable information source for state-of-the-art research and developments in machine learning for mobile edge computing and wireless mobile networks.It also aims to serve as an outlet for facilitating computational intelligence among mobile edge computing researchers,practitioners,and professionals across academia,government and industry.Finally,it aims to foster the dissemination of high-quality research on new ideas,methods,theories,techniques,and applications of evaluation and management for improving mobile services.Original research articles are solicited in all aspects,including theoretical studies,practical applications,and experimental prototypes.The special issue is composed of seven papers organized as follows.

The first work is entitled ‘Customer Tiered Purchase Forecast by mobile edge computing based on Pareto/NBD and SVR’.It proposes a ARIMA model to predict the overall sales volume of the enterprise,and then applies the SVR model and the Pareto/NBD model into real situation to make layered predictions of the arrival of new and old customers.The authors further provide advice for the enterprise to divide the customer structure and then do marketing prediction more accurately.

The second work is entitled ‘Multi-objective Task Assignment for Maximizing Social Welfare in Spatio-temporal Crowdsourcing’.It combines Spatio-temporal crowdsourcing and edge computing to study the Multi-Objective Optimization Task Assignment (MOO-TA) problem in the edge computing environment.The proposed online incentive mechanism considers the task difficulty attribute to motivate crowd workers to perform sensing tasks in the unpopular area.In addition,it proposes the Weighted and Multi-Objective Particle Swarm Combination (WAMOPSC) algorithm is to maximize both platform’s and crowd workers’ utility,so as to maximize social welfare.

The third work is entitled ‘Deep Reinforcement Learning-based Computation Offloading for 5G Vehicle-aware Multi-access Edge Computing Network’.It proposes a 5G-based vehicle-aware Multi-access Edge Computing network (VAMECN)and a joint optimization problem of minimizing total system cost.To solve the problem,it uses a deep reinforcement learning-based joint computation offloading and task migration optimization (JCOTM)algorithm,which considers the influences of multiple factors such as concurrent multiple computation tasks,system computing resources distribution,and network communication bandwidth.

The fourth work is entitled ‘Privacy-Preserving Collaborative Filtering Algorithm Based on Local Differential Privacy’.It designs a recommendation framework that applies local differential privacy(LDP) to collaborative filtering.In the proposed framework,users’ rating data are perturbed to satisfy LDP and then released to the server.The server computes the similarity between items by using the perturbed data.It also proposes a data perturbation method to protect user’s original rating values,where the Harmony mechanism is modified so as to preserve the accuracy of similarity computation.

The fifth work entitled ‘Joint Design of Content Delivery and Recommendation in Wireless Caching Networks’ proposes a joint transmission scheme of content objects and recommendation lists with edge caching,and formulates an optimization problem to balance the utility and cost of content caching and recommendation,which is a mixed integer nonlinear programming problem.Then,it proposes a reinforcement learning based algorithm to implement real time management of content caching,recommendation and delivery,which can approach the optimal solution without iterations during each decision epoch.

The sixth work entitled ‘Remaining Time Prediction for Business Processes with Concurrency Based on Log Representation’ proposes a new method to predict the remaining time for business processes based on trace representation is proposed.More specifically,it first associates the prefix set generated by the event log to different states of the transition system,and encodes the structural features of the prefixes in the state.Then,an annotation containing the feature representation for the prefix and the corresponding remaining time are added to each state to obtain an extended transition system.Next,states in the extended transition system are partitioned by the different lengths of the states,which considers concurrency among multiple process instances.Finally,the Long Short-Term Memory (LSTM) deep recurrent neural networks are applied to each partition for predicting the remaining time of new running instances.

Finally,the seventh work,entitled ‘Novel Private Data Access Control Scheme Suitable for Mobile Edge Computing’,proposes an anonymous privacy data protection and access control scheme based on elliptic curve cryptography (ECC) and bilinear pairing to protect the communication security of the MEC.In the proposed scheme,the information sender encrypts private information through the ECC algorithm,and the information receiver uses its own key information and bilinear pairing to extract and verify the identity of the information sender.During each round of communication,the proposed scheme uses timestamps and random numbers to ensure the freshness of each round of conversation.Experimental results show that the proposed scheme has good security performance and can provide data privacy protection,integrity verification,and traceability for the communication process of MEC.

To summarize,this special issue has collected top papers from the main institutions,projects and standardization bodies working on the topic of machine learning on mobile edge computing.These works present a bottom-up approach covering from the fundaments based on its integration.

The guest editors would like to thank the authors for their contribution and the reviewers for their great efforts to provide insightful and valuable reviews.They hope that the ongoing research in this special section will provide further insight.