Theoretical design of decentralized auction framework under mobile crowdsourcing environment

Highlights

• We enriched the background in the introduction section.

• We added a lot of the latest references.

• We refined the language and descriptions to make it more clear.

• We added many comparative experiments and figures.

Abstract

With the rapid popularization of mobile devices, the mobile crowdsourcing has become a hot topic in order to make full use of the resources of mobile devices. To achieve this goal, it is necessary to design an excellent incentive mechanism to encourage more mobile users to actively undertake crowdsourcing tasks, so as to achieve maximization of certain economic indicators. However, most of the reported incentive mechanisms in the existing literature adopt a centralized platform, which collects the bidding information from workers and task requesters. There is a risk of privacy exposure. In this paper, we design a decentralized auction framework where mobile workers are sellers and task requesters are buyers. This requires each participant to make its own local and independent decision, thereby avoiding centralized processing of task allocation and pricing. Both of them aim to maximize their utilities under the budget constraint. We theoretically prove that our proposed framework is individual rational, budget balanced, truthful, and computationally efficient, and then we conduct a group of numerical simulations to demonstrate its correctness and effectiveness.

Introduction

During the last ten years, mobile devices have been getting stronger and stronger by installing multiple sensors, microcomputers, and communication equipments. It forms a new pattern, mobile crowdsourcing (MC), which has attracted wide attention from academia and industry because of its great commercial value. The MC refers to the use of computing resources or sensory abilities of a group of mobile users to accomplish different types of tasks. A lot of different applications based on MC are gradually being industrialized, such as traffic monitoring, healthcare, and electric vehicle charging. As for a crowdsourcing platform, it wants as many mobile users as possible to participate in the crowdsourcing task. However, mobile users are reluctant to do so, because it not only consumes energy and time, but also risks exposing privacy. Therefore, while ensuring security, it is a core issue to promote users' participation.

In the existing works, a great deal of research has been done to promote users' participation in crowdsourcing by designing incentive mechanisms [9] [4]. Auction theory [19] [31] is commonly used in incentive mechanisms for its advantage in dealing with the interaction between buyers and sellers. In MC, it can be exploited to determine the hired mobile users and the fees paid to them, in which mobile workers are sellers and task requesters are buyers. Nevertheless, there are several severe issues in the existing auction mechanisms [31] [24] [33] [10]. First, the complexity and generalization of both tasks and workers are not fully studied, which can be shown as follows.

• A task is published by a requester, and the task can be undertaken by many workers, but the total remuneration paid by the requester for the task is limited;

• A worker can take on several different tasks at the same time, but the resource and energy investment it puts into each tasks are different.

• For each worker, its total investment to tasks is limited.

• For the same investment, different workers bear the cost of it differently.

Second, in the existing auction design, they always assume that there is a fair and right-minded platform/auctioneer to collect the bidding information from buyers and sellers, and then determine the auction results. Such a centralized management pattern is not conducive to managing the joining and leaving of requesters and workers in a dynamic environment, but also threatens the users' privacy, thus reducing the incentive effect. Taking these two issues into consideration, it is a challenge to incentivize both requesters and workers that adapts to complexity and privacy protection simultaneously.

Based on the above careful observation, we propose a decentralized auction framework (DAF), and take DAF as an incentive mechanism to achieve the assignment and pricing between tasks and workers. The architectures of centralized and decentralized MC system can be shown in Fig. 1. Here, we consider the flexibility of real crowdsourcing scenarios as much as possible, and generalize it to a concise mathematical expression, which mainly covering the following points:

• Budget constraints: the total payment paid to workers for each task has been constrained by the requester and the total investment in the task has been constrained by the worker.

• For each worker, it can offer a arbitrary investment with its own bid according to its cost.

• Our proposed DAF can be carried out in a decentralized manner, in which each requester can determine the winning workers to complete its task with the corresponding payment and each worker can determine whether it is willing to undertake the tasks locally and independently.

In such a decentralized mechanism, it can be further integrated with emerging technologies (such as distributed computing, edge computing, and blockchain). Take the blockchain as an example, its consensus process can be merged into our auction framework to achieve distributed storage and system's security. At the same time, our DAF satisfies several design rationales of auction theory: individual rationality, budget balance, truthfulness, and computational efficiency. This will avoid price manipulation and guarantees a fair and competitive market environment. To our best knowledge, this is the first time to put forward a decentralized and truthful auction mechanism like us to address the MC problem. Thus, this is our main contribution. For convenience, we will use buyers and sellers interchangeably with task requesters and mobile users (workers) in the rest of this paper. Finally, we conduct intensive simulations to evaluate the performance of our proposed algorithms, whose results verify the correctness and effectiveness of our theoretical analysis.

Organization: In Section 2, we summarize the works related to this paper. We then introduce our model and MC problem in Section 3, and algorithm design in Section 4. In Section 5 and Section 6, we conduct the theoretical analysis for our auction framework and experiment to validate our algorithm. Finally, Section 7 concludes this paper.