Elsevier

Computer Networks

Volume 54, Issue 4, 19 March 2010, Pages 589-604
Computer Networks

Design and performance evaluation of ContentPlace, a social-aware data dissemination system for opportunistic networks

https://doi.org/10.1016/j.comnet.2009.09.001Get rights and content

Abstract

In this paper we present and evaluate ContentPlace, a data dissemination system for opportunistic networks, i.e., mobile networks in which stable simultaneous multi-hop paths between communication endpoints cannot be provided. We consider a scenario in which users both produce and consume data objects. ContentPlace takes care of moving and replicating data objects in the network such that interested users receive them despite possible long disconnections, partitions, etc. Thanks to ContentPlace, data producers and consumers are completely decoupled, and might be never connected to the network at the same point in time. The key feature of ContentPlace is learning and exploiting information about the social behaviour of the users to drive the data dissemination process. This allows ContentPlace to be more efficient both in terms of data delivery and in terms of resource usage with respect to reference alternative solutions. The performance of ContentPlace is thoroughly investigated both through simulation and analytical models.

Introduction

This paper is focused on the problem of data dissemination in opportunistic networks. Opportunistic networks [18] are a recent mobile networking paradigm stemming from the research on conventional Mobile Ad Hoc NETworks (MANET). In this paradigm nodes are assumed to be mobile, and forwarding of messages occurs based on the store-carry-and forward concept [9]. Specifically, no simultaneous end-to-end multi-hop path is required to enable communication between any two nodes, unlike conventional MANET. Instead, each node carrying a message for an intended destination evaluates the suitability of any other node it makes contact with as the next hop. Messages are thus opportunistically forwarded by exploiting nodes encounters, until they reach the intended destination. This paradigm enables end-to-end communication despite possible long disconnections of the communication endpoints and severe network partitions, which is usually cumbersome in traditional MANET architectures.

A significant share of research on opportunistic networks has focused on routing issues (see [18] for a survey). Instead, in this paper, we consider the problem of data dissemination. This is a key research problem, particularly in opportunistic networks. In this environment, according to the user-generated content wave, users are expected to generate large amounts of content by exploiting capability-rich mobile devices (such as PDAs, smartphones, etc.), and to share them with people around them or people they have social relationships with. In the following, we refer to any piece of content (e.g., a picture or an mp3 file) as a data object. The problem of efficiently disseminating data objects in opportunistic networks is thus very relevant, and not widely explored in the literature yet, as discussed in Section 2.

Data dissemination in opportunistic networks is a difficult problem. As the topology is very unstable, and users appear in and disappear from the network dynamically, content providers and content consumers might be completely unaware of each other, and never connected at the same time to the same part of the network. Therefore, data objects should be moved and replicated in the network in order to carry them to interested users despite disconnections and partitions. On the other hand, data dissemination systems should take care of both network and device resource constraints. For example, a trivial solution would be to flood the whole network with any generated data object, but this would clearly saturate both network resources (in terms of available bandwidth) and device resources (e.g., in terms of energy, storage, etc.).

In this paper, we propose and evaluate ContentPlace, which is a data dissemination system for opportunistic networks that exploits social information about users behaviour in order to drive the dissemination process. Exploiting social information is a very promising research direction for opportunistic networks. In this environment the nodes are mobile devices users carry with them all the time. Therefore, the users social behaviour, being a key driver for their movement patterns, is also a key piece of context information to predict nodes’ co-location and future encounters. ContentPlace assumes that users belong to social communities, and autonomically learns the time spent by them in each community, which types of data objects users of each community are interested in, and how spread in the communities the data objects are. This information is used to evaluate the utility of each encountered data object. Specifically, each node, upon making contact with another peer, evaluates the utility of the data objects the peer is carrying. Assuming that the buffer space devoted to the dissemination process is limited, the node selects which data objects to fetch from the peer, in order to maximise the total utility of the data objects in its own buffer. Therefore, the data dissemination process is driven by the interests and social behaviour of the users, and just requires local interactions between nodes that happen to come in contact.

We assume that users’ mobility is driven by the social relationships among users, i.e., that users spend their time with their friends. This assumption may not be always satisfied. This is the case, e.g., of friends living in different part of the world or of virtual friends (e.g., Facebook or chatroom friends). However, also these relationships can be exploited to disseminate messages. For example, in the case of long distance friendship, the distance can be overcome by going through the traditional infrastructure and then switching again to the ad hoc mode when this distance has been covered. The inclusion of such hybrid communications into our data dissemination system is currently under study.

After presenting ContentPlace in Sections 3 ContentPlace general design, 4 Contentplace social design, we provide a detailed simulation analysis in Section 5. Specifically, we compare different data dissemination policies that could be plugged in the general ContentPlace design, either considering or not considering social information. We show that social-aware policies outperform naïve ones in which social information is not taken into account. Among the social-aware policies, we identify the one performing best (named Future), and highlight the reasons why it is the most efficient one. Finally, in order to shed additional light on the performance of the Future policy, in Section 6 we provide an analytical model describing its behaviour.

Section snippets

Related work

Content dissemination systems have been proposed with regard to legacy Internet networks [1], and also with respect to conventional MANETs [22]. In general, these systems assume that network paths are rather stable, and in some cases generate a significant amount of traffic to maintain knowledge of other nodes’ caches. Therefore, they are not suitable to opportunistic networks.

The idea behind ContentPlace is to exploit social information on the environment the nodes operate in, in order to

ContentPlace general design

This section provides necessary background information required to present the main contribution of this paper by recalling the target application scenario and the main design features of the ContentPlace system.

Contentplace social design

The use of the social weights in Eq. (2) permits full flexibility in the ContentPlace design. Specifically, it allows us to define and compare different social-oriented policies, as well as, in the limiting case, non-social policies such as a greedy behaviour. Although clearly not exhaustive, the set of social-oriented policies we compare covers a fairly large spectrum of possible (reasonable) definitions. Overall, the global goal we wish to achieve is to optimise the hit rate for all users in

Performance evaluation

In this section, we evaluate the performance of the social-oriented policies described in Section 4. To this aim, we developed a custom simulator that is extended from the one in [6] and uses the same assumptions for lower communication layers. As we discuss in the following, the simulation scenario we consider has been chosen because it is able to highlight general features of the social-oriented policies. Note that in [5] we have already presented results showing the impact of system

Average analysis of the Future policy

In this Section we perform an average analysis of the dissemination behaviour of the Future policy, that we have shown to be the best in Section 5. The scenario we consider is that of two communities A and B, having N nodes each that roam locally in the community, with a traveller node commuting between A and B. We consider the following situation: (i) an initial set of data has been already spread among the nodes of the two communities and its dissemination process has ended, leaving the

Conclusions

In this paper we have focused on data dissemination issues for opportunistic networks. Specifically, we have presented and evaluated ContentPlace, which is a data dissemination system exploiting information about the social behaviour of the users in order to drive the dissemination process. We have provided extensive simulation results showing the advantage of social-aware policies over naïve policies that do not take social information into account. In general, social-aware policies turn out

Chiara Boldrini is a Ph.D. candidate at the IIT Institute of the Italian National Research Council (CNR), Italy. She holds a M.Sc. degree (2006) in Computer Engineering from the University of Pisa. She is working on social-aware data dissemination, routing protocols, and mobility models for opportunistic networks.

References (23)

  • R. Groenevelt et al.

    The message delay in mobile ad hoc networks

    Perform. Evaluat.

    (2005)
  • A. Balamash et al.

    An overview of web caching replacement algorithms

    IEEE Commun. Surv. Tut.

    (2004)
  • A. Balasubramanian, B.N. Levine, A. Venkataramani, Dtn routing as a resource allocation problem, in: ACM SIGCOMM,...
  • C. Boldrini, M. Conti, A. Passarella, Users mobility models for opportunistic networks: the role of physical locations,...
  • C. Boldrini, M. Conti, A. Passarella, ContentPlace: social-aware data dissemination in opportunistic networks, in: ACM...
  • C. Boldrini, M. Conti, A. Passarella, Context and resource awareness in opportunistic network data dissemination, in:...
  • C. Boldrini et al.

    Exploiting users’ social relations to forward data in opportunistic networks: the HiBOp solution

    Pervasive Mob. Comput.

    (2008)
  • C. Boldrini, M. Conti, A. Passarella, IIT TR-08/2009, Technical Report, IIT-CNR, 2009....
  • P. Costa et al.

    Socially-aware routing for publish-subscribe in delay-tolerant mobile ad hoc networks

    IEEE JSAC

    (2008)
  • K. Fall, A delay-tolerant network architecture for challenged internets, in: ACM SIGCOMM, 2003, pp....
  • P. Hui, J. Crowcroft, How small labels create big improvements, in: IEEE ICMAN,...
  • Cited by (86)

    • The Internet of People (IoP): A new wave in pervasive mobile computing

      2017, Pervasive and Mobile Computing
      Citation Excerpt :

      Indeed, opportunistic networks mimic, in the cyber world, the way information disseminates in the physical world through the contacts among people. To achieve this, personal devices should contribute part of their resources (e.g., energy, computing and storage capabilities) to the dissemination process [189–197]. Thus, the data selection process must be very lightweight and able to perform a sharp distinction between data items, since only a very limited portion of them can be stored.

    • Information-Centric Networking in next-generation communications scenarios

      2017, Journal of Network and Computer Applications
    View all citing articles on Scopus

    Chiara Boldrini is a Ph.D. candidate at the IIT Institute of the Italian National Research Council (CNR), Italy. She holds a M.Sc. degree (2006) in Computer Engineering from the University of Pisa. She is working on social-aware data dissemination, routing protocols, and mobility models for opportunistic networks.

    Marco Conti is a research director at IIT, an institute of the Italian National Research Council (CNR). He published in journals and conference proceedings more than 200 research papers related to design, modelling, and performance evaluation of computer-network architectures and protocols. He co-authored the book “Metropolitan Area Networks” (1997) and is co-editor of the books “Mobile Ad Hoc Networking” (2004) and “Mobile Ad Hoc Networks: From Theory to Reality” (2007). He is the chair of the IFIP WG 6.3 “Performance of Communication Systems”. He is the Editor-in-chief of the Computer Communications Journal and Associate Editor-in-chief of Pervasive and Mobile Computing Journal; he is on the editorial board of IEEE Transactions on Mobile Computing, Ad Hoc Networks, Journal of Communications Systems, and Wireless Ad Hoc and Sensor Networks: An International Journal. He served as TPC chair of IEEE PerCom 2006, and of the IFIP-TC6 Conferences Networking 2002 and PWC 2003, and as TPC Co-chair of ACM WoWMoM 2002, WiOpt’04, IEEE WoWMoM 2005, and ACM MobiHoc 2006. He served as general chair of ACM REALMAN 2006 and IEEE MASS 2007, and as general Co-chair of IEEE WoWMoM 2006, and ACM MobiOpp 2007. He is currently serving as general Co-chair of IEEE PerCom 2010 and as program Co-chair of ACM MobiOpp 2010.

    Andrea Passarella is with IIT-CNR, Italy. Previously, he was a Researcher at the Computer Laboratory, Cambridge, UK. He holds a Ph.D. in Computer Engineering (Pisa University, Italy). He is working on opportunistic self-organising networks, with emphasis on p2p, services, routing protocols, mobility models. He serves in the Program Committes, among others, of IEEE WoWMoM and PerCom. He was Co-chair of IEEE AOC 2009, TPC Co-chair of ACM MobiOpp 2007, Vice-chair for ACM REALMAN (2005/2006), and IEEE MDC (2006). He is Workshops Co-chair for IEEE PerCom and WoWMom 2010. He is an Associate Technical Editor for IEEE Communications Magazine, and is in the Editorial Board of the Inderscience IJAACS Journal.

    View full text