Task Scheduling For Enabling Ubiquitous Sensing Service In Mobile Edge Computing

With the development of Internet of Things(IoT)and mobile technologies,the number of smart connected devices,including mobile devices,sensors and actuators,is growing rapidly.These devices not only consume the data but also collect massive data.At the same time,the rise of IoT has also promoted a new network model,which makes the connection of all things more extensive and close without the traditional Internet technologies.Edge computing has emerged when traditional cloud computing cannot meet the growing demand for data computing and services.Edge computing refers to data processing at the edge of the network close to objects or data generation,thereby effectively reducing the resource consumption of data transmission,network bandwidth,data backup,etc..Therefore,it adapts to the various Internet scenario better.In real-world scenarios,in addition to providing computing and other specific services for edge devices,mobile edge servers play an role in sensing data collecting.Then we consider sensing tasks in edge computing.On the one hand,edge devices and edge servers are used to share the computing tasks of the cloud server,reducing the amount of data transmission and consume less network bandwidth;on the other hand,mobility of edge devices and instability of the network makes it impossible to keep task allocation static.It has to adapt to the dynamic network topology,taking into account the resources,computing power,and server availability of the edge device itself.This paper will propose a dynamic sensing task scheduling strategy to meet the multi-faceted goals of energy consumption of edge devices and response time of tasks.This paper first presents the problem we are studying,and then introduces the hierarchy of the mobile edge computing framework,including three parts: edge device layer,edge cloud layer and central cloud layer.The mobile edge device is the source of the sensing data.The edge cloud is deployed in the wireless access network to provide data collection services.The local controller is responsible for the distribution,maintenance and monitoring of the perceptual tasks,and the central cloud layer is responsible for the service registration and data results.The architecture leverages the sensing capabilities,computing power,and computing abilities of mobile edge devices to reduce the power consumption as well as effectively reduce delay.It is important to model edge devices and service to abstract resources and provide a unified formal description of service for various sensing data.In addition to the sensing capabilities of the device,the speed and position information of the device is the basis of our dynamic scheduling decision.We have a unified depiction of all relevant attributes,giving the model of the device and the model of the task.Users customize the service by pre-defined properties.In order to achieve dynamic perceptual task scheduling,we first need to consider the goal of scheduling optimization: the energy consumption and delay of the task;then we propose the energy consumption and delay model,and define the target model.Then we proposed the concept of virtual sensor,and formalized the problem as a minimum integer programming problem.It is transformed into the maximum matching of the weighted bipartite graph and solved by two steps: i)initial assignment of tasks,implemented by greedy algorithm;ii)maintained by using the 2-approximation algorithm,making the update complexity time is O(log n)per side.In the perceived service,we define SLA of the service to measure the quality.To improve the availability of services,the combination of proactive and reactive recovery is adopted when any server failures.It can not only utilize the reactive recovery model to save redundant resources,but also play an active recovery model to shorten the failure time.As for the proactive recovery model,the biggest challenge is how to determine the number and location of secondary backups,which is solved as a collection coverage problem.Finally,this paper constructs a prototype of the perceptual task dynamic scheduling scheme in the mobile edge computing environment.The simulation experiments are carried out under the different weight of cost model factors,and compared with the dynamic stochastic dynamic task scheduling algorithm.The experimental results show that the dynamic scheduling mechanism of sensing task in the mobile edge computing comprehensively considers the energy consumption,the delay time and the mobility of devices,and realizes the dynamic task adjustment strategy.