A location-constrained crowdsensing task allocation method for improving user satisfaction

Huihui Chen,Liming Chen,Zhiwen Yu,Bin Guo

doi:10.1177/1550147719883987

Abstract

Mobile crowdsensing is a special data collection manner which collects data by smart phones taken by people every day. It is essential to pick suitable workers for different outdoor tasks. Constrained by participants’ locations and their daily travel rules, they are likely to accomplish light outdoor tasks by their way without extra detours. Previous researchers found that people prefer to accomplish a certain number of tasks at a time; thus, we focus on assigning light outdoor tasks to workers by considering two optimization objectives, including (1) maximizing the ratio of light tasks for different workers and (2) maximizing the worker’s satisfaction on assigned tasks. This task allocation problem is a non-deterministic polynomial-time-hard due to two reasons, that is, tasks and workers are many-to-many relationships and workers move from different places to different places. Considering both optimization objectives, we design the global-optimizing task allocation algorithm, which greedily selects the most appropriate participant until either no participant can be chosen or no tasks can be assigned. For the purpose of emulating real scenarios, different scales of maps, tasks, and workers are simulated to evaluate algorithms. Experimental results verify that the proposed global-optimizing method outperforms baselines on both maximization objectives.

Highlights

The smart phones, wearable devices, and mobile social networking techniques have made mobile crowdsensing (MCS) and computing (MCSC)1 a popular research area in recent years
For the purpose of both greedily assigning more tasks and satisfying most workers, we propose worker-first globally optimized (WF-GO) algorithm, which searches the most suitable worker rather than assigning the sequential candidate proper tasks like workerfirst local-optimizing (WF-LO)
In order to evaluate the performance of the WF-GO method when the candidate number is small, we increase the task number but stabilize the worker number

Summary

Introduction

The smart phones, wearable devices, and mobile social networking techniques have made mobile crowdsensing (MCS) and computing (MCSC) a popular research area in recent years. The MCS platform computes personalized suitable tasks for those participants according to their queries by using proposed task allocation methods in this article and uses the push-based method to publish suitable tasks to participants who are chosen to be workers. To address the second challenge, we utilize the stationary locations of tasks to create the light task relationship matrix for every worker candidate and globally optimize workers’ tasks with this matrix. We propose a global-optimizing task allocation algorithm, and the participant who can take the maximum suitable tasks will be prior chosen to be a worker In this way, tasks are allocated as many as possible and workers will get their expected number of tasks leading to satisfied experience with task allocation. Section ‘‘Evaluation’’ presents experiments and discuss the results, followed by conclusions in section ‘‘Conclusion.’’

Related work

Evaluation

Experimental results

Conclusion