Please use this identifier to cite or link to this item: http://hdl.handle.net/1783.1/79500

OPENRP: A Reputation Middleware for Opportunistic Crowd Computin

Authors Chatzopoulos, Dimitrios HKUST affiliated (currently or previously)
Ahmadi, Mahdieh HKUST affiliated (currently or previously).
Kosta, Sokol
Hui, Pan View this author's profile
Issue Date 2016
Source IEEE Communications Magazine , v. 54, (7), July 2016, article number 7509388, p. 115-121
Summary The concepts of wisdom of crowd and collective intelligence have been utilized by mobile application developers to achieve large-scale distributed computation, known as crowd computing. The profitability of this method heavily depends on users' social interactions and their willingness to share resources. Thus, different crowd computing applications need to adopt mechanisms that motivate peers to collaborate and defray the costs of participating ones who share their resources. In this article, we propose OPENRP, a novel, lightweight, and scalable system middleware that provides a unified interface to crowd computing and opportunistic networking applications. When an application wants to perform a device-to-device task, it delegates the task to the middleware, which takes care of choosing the best peers with whom to collaborate and sending the task to these peers. OPENRP evaluates and updates the reputation of participating peers based on their mutual opportunistic interactions. To show the benefits of the middleware, we simulated the behavior of two representative crowdsourcing applications: message forwarding and task offloading. Through extensive simulations on real human mobility traces, we show that the traffic generated by the applications is lower compared to two benchmark strategies. As a consequence, we show that when using our middleware, the energy consumed by the nodes is reduced. Finally, we show that when dividing the nodes into selfish and altruistic, the reputation scores of the altruistic peers increase with time, while those of the selfish ones decrease.
ISSN 0163-6804
Language English
Format Article
Access View full-text via DOI
View full-text via Scopus
View full-text via Web of Science
Find@HKUST