JDER: A history-based forwarding scheme for delay tolerant networks using Jaccard distance and encountered ration

https://doi.org/10.1016/j.jnca.2013.09.012Get rights and content

Abstract

Delay tolerant networks have arisen as a new paradigm of wireless communications in which nodes follow a store-carry-and-forward operation. Unlike other ad hoc networks, mobility of nodes is seen as an interesting feature to deliver information from a source node to a destination node. New forwarding schemes have been proposed to deal with the intermittent communications carried out by nodes in delay tolerant networks. Most forwarding schemes assume that nodes are divided into social communities and the communications are likely to be established between two nodes belonging to the same community. However, the social information is not always available, especially in large environments like cities so it has to be inferred from the history of encounters among nodes. Furthermore, there are cases in which the information has to be widely disseminated throughout the network such as alarm and emergency messages so it has to pass through different communities. In this paper, we propose JDER, a new probabilistic forwarding scheme which guarantees high reachability throughout the network by selecting cut-nodes. JDER is based on two metrics: the history encountered ration and the Jaccard distance, and it has been extensively validated through simulations using 8 different mobility models obtained from real life traces.

Introduction

Delay Tolerant Networks (DTNs) are decentralized networks in which nodes cooperatively communicate to transmit application data from a source node to a destination node. Unlike other ad hoc networks such as Mobile Ad Hoc Networks (MANETs) (Hoebeke et al., 2004) and Wireless Sensor Networks (WSNs) (Akyldiz et al., 2002), network partitioning, disconnections, and high end-to-end delay are dominant factors in DTNs due to dynamic topologies, mobility of nodes and low density conditions. While in MANETs and WSNs mobility of nodes causes undesirable effects such as broken links (Gutiérrez-Reina et al., 2011, Gutiérrez-Reina et al., 2012a) and low delivery rates deteriorating the performance of such networks, in DTNs mobility is used as an opportunity to deliver information. DTNs are suitable for a large range of applications including disaster scenarios (Martín-Campillo et al., 2013), intelligent transportation systems (Gutiérrez-Reina et al., 2012b, Shin et al., 2012), and pervasive healthcare, among others (Conti and Kumar, 2010b). In DTNs, nodes follow a store-carry and-forward operation (Pelusi et al., 2006, Conti et al., 2010a). Whenever a given node (source node) has certain information to be transmitted to a destination node, it will opportunistically transmit a copy of such information to one or several intermediate nodes. These intermediate nodes will carry and forward the information again to new intermediate nodes until a copy of the message reaches the destination node. Therefore, the forwarding schemes are responsible for deciding which nodes have to retransmit the information. Consequently, the main goal of the forwarding schemes is to select suitable intermediate nodes that are likely to communicate with the target destination node. Many aspects can be considered to efficiently select a forwarding criterion such as geographical information (Leontiadis and Mascolo, 2007), social information (Boldrini et al., 2010a, Gao et al., 2012), among others. If geographical information is used, nodes need to be equipped with a positioning system like a Global Positioning Systems (GPS) incurring additional cost. However due to the significant variation in the location of both the source and the destination nodes over time, the positioning information becomes inefficient in DTNs. On the other hand, if social information is used, nodes have to be aware of the social relationships among nodes in the network. However, the availability of social information may not always be possible due to the size of the network and/or the lack of access to a real life social network like Facebook or Twitter. In order to overcome the above mentioned drawbacks, forwarding schemes based on the history of encounters among nodes in the networks have been proposed (Boldrini et al., 2007, Lindgren et al., 2003). These are aimed to use the past history of encounters among nodes and to select the ones that are likely to communicate with the destination nodes thus restricting the number of retransmissions to low values. The main idea is to predict future encounters among nodes using the past history of encounters, preferences and similarities among nodes (Boldrini et al., 2007, Ciobanu et al., 2013). However, prediction mechanisms have the drawback of being hard to tune successfully depending on the specific behavior of the network, especially when dealing with highly dynamic DTNs. In contrast to the previously proposed algorithms for DTNs, we specifically focus on using the properties of cut-nodes (cut-vertex in the social networks theory) in order to make retransmission decisions. Cut-nodes are defined as those nodes whose deletion increases the number of components in the network (De Nooy et al., 2005), and they occupy critical positions for the flow of information within the network as they control the flow from one part to another part of the network. The main advantage of using cut nodes as forwarding nodes in DTNs is that they belong to more than one community so they are vital to message retransmission whether the destination node belongs to the same community or different community to the source node. Therefore, considering the cut-nodes as forwarders will guarantee a high delivery ratio in DTNs. Though identifying cut-nodes may be a trivial task when the whole network is known, this is not the case in DTNs due to their dynamic topology and low connectivity of nodes. In this paper, we propose a method for identifying such nodes by combining two different measures: (1) the history of encountered ration, which finds possible forwarding nodes belonging to the same social network, and (2) the Jaccard distance (explained in more detail in Section 3) between two nodes, which measures the dissimilarity between two nodes (Gutiérrez-Reina et al., 2013). By combining these two measures, nodes can identify possible forwarding nodes that belong to the same social network and at the same time, are connected to nodes belonging to other communities.

The main contributions of this paper are

  • To propose a new probabilistic forwarding scheme based on cut nodes for DTNs.

  • To evaluate the proposed forwarding scheme using real-life mobility traces.

  • To compare the performance of the proposed scheme with that of other forwarding schemes found in the literature.

This paper continues as follows. A brief review of existing forwarding schemes for DTNs is presented in Section 2. Section 3 discusses the characteristics of cut-nodes in detail. The proposed forwarding scheme is described in Section 4 along with the procedure to calculate the history of encountered ration using the history of encounters among nodes and the Jaccard distance using neighborhood information. Section 5 presents information on how the simulations were conducted and a discussion of the simulation results. Finally, Section 6 includes the main conclusions of this paper.

Section snippets

Forwarding schemes in delay tolerant networks

Since mobile devices have become very popular in recent years, DTNs have been the focus of increasing research. The major issues in DTNs concern routing (Tahsin et al., 2011), forwarding and dissemination (Conti et al., 2010a) of information.

A taxonomy for opportunistic data dissemination algorithms has been proposed in Ciobanu and Dobre (2011) and describes such techniques using four main categories: network infrastructure (how the network is organized using overlays for nodes), node

Finding cut-nodes in social networks

Cut-nodes play an important role in the connectivity of different communities in social networks. If cut-nodes are not taken into account by the forwarding schemes, some nodes in the network will never be reached. Let us consider the following example, Fig. 1, to illustrate the importance of cut-nodes in the dissemination process.

In the above Fig. 1 we can distinguish three different communities (circles, squares, and hexagons). The gray nodes represent the cut-nodes and they link different

JDER (Jaccard distance-encountered ration)

To achieve a high delivery ratio in DTNs with a low delivery cost, we propose JDER, a novel probabilistic forwarding scheme. The main objectives of JDER are

  • To select nodes inside the same social network by using the history encountered ration.

  • To select dissimilar nodes inside the same community by using the Jaccard distance. It will enable the exploration of new regions of the network through cut-nodes as well as the reduction of the delivery cost for transmitting information inside the same

Experimental setup and simulation results

This section presents an experimental analysis of the JDER data dissemination algorithm for DTNs, in terms of five metrics chosen to highlight various capabilities of such algorithms. We compare the performance of the algorithm presented above to that of distributed BUBBLE Rap (Hui et al., 2008, Hui et al., 2011), which is one of the most well-known and efficient data dissemination algorithms in terms of hit rate and delivery latency, and SPRINT (Ciobanu et al., 2013), a socially-aware

Conclusions

There are several conclusions that can be drawn from the results presented in the previous section. First of all, we have shown that an algorithm based on the Jaccard distance can be used for routing in opportunistic networks. Generally, for all cases we tested, the hit rate obtained is improved when using JDER. This happens because the decision of a message's next hop is more informed, and thus it has a greater chance of being delivered.

In terms of congestion, we obtain good results (which are

Acknowledgments

This work was supported in part by the University of Seville under the PhD grant PIF (Personal Investigador en Formación) of Daniel Gutiérrez Reina. This paper has also benefited from the collaborative research efforts of the EU Green-Net group.

References (36)

  • Ciobanu RI, and Dobre C. Data dissemination in opportunistic networks. In: Proceedings of the 18th international...
  • Ciobanu RI, Dobre C, Cristea V, Al-Jumeily D. Social opportunistic networking. In: Proceedings of 11th international...
  • Ciobanu RI, Dobre C, Cristea V. Social aspects to support opportunistic networks in an academic environment. In:...
  • Ciobanu RI, Dobre C, Cristea V. Reducing congestion for routing algorithms in opportunistic networks with...
  • M. Conti et al.

    From opportunistic networks to opportunistic computing

    IEEE Communications Magazine

    (2010)
  • M. Conti et al.

    Opportunities in opportunistic computing

    IEEE Computer Magazine

    (2010)
  • CRAWDAD....
  • W. De Nooy et al.

    Exploratory network analysis with Pajek (structural analysis in the social sciences)

    (2005)
  • Cited by (19)

    • General and mixed linear regressions to estimate inter-contact times and contact duration in opportunistic networks

      2019, Ad Hoc Networks
      Citation Excerpt :

      Based on this contact prediction model, along with other social criteria, the authors propose in [23] a new opportunistic routing algorithm (SPRINT), which improves the delivery ratio of messages as compared to traditional social-based routing approaches. A further improvement is JDER [24], a new probabilistic forwarding scheme also based on the history of previously encountered nodes, one that determines the network cut-nodes, which will increase the probability of reaching the destination nodes. Another approach is to use a multi-objective model and decision tree based mechanism for optimising data dissemination under different target performances [25].

    • A survey on unmanned aerial and aquatic vehicle multi-hop networks: Wireless communications, evaluation tools and applications

      2018, Computer Communications
      Citation Excerpt :

      MANETs were conceived originally as a way to connect wirelessly different portable devices such as smartphones, laptops, and tablets located close to each other. As an evolution of MANETs, Delay Tolerant Networks (DTNs) [23] are low density networks with very high mobility. DTNs aim to transmit data packets in disconnection-prone environments and with non-guaranteed end-to-end connections.

    • MATEM: A unified framework based on trust and MCDM for assuring security, reliability and QoS in DTN routing

      2018, Journal of Network and Computer Applications
      Citation Excerpt :

      A single copy of each message is forwarded to the qualified node. Most of these knowledge-based protocols select a suitable next-hop carrier based on contact history of potential carriers (Ciobanu et al., 2014a; Ayub et al., 2014), knowledge about traffic patterns in the network (Shin et al., 2012) or on probability of encountering the destination node (Yuan et al., 2012). Furthermore, some of them have used multi-copy spraying mechanisms to improve reliability amidst intermittent connectivity (Bulut et al., 2010; Niu et al., 2015).

    • Real experimentation of probabilistic broadcasting algorithms based on dissimilarity metrics for multi-hop ad hoc networks

      2016, Ad Hoc Networks
      Citation Excerpt :

      This is a mandatory condition, since in wireless multi-hop networks nodes may only have local information. The Jaccard distance has been already used in mobile ad hoc networks [28] to improve the discovery phase of the AODV routing protocol [31] and in opportunistic networks to enhance data dissemination [32]. This coefficient is not very different in form from the Jaccard coefficient.

    • Routing Protocols in an Opportunistic Network: A Survey

      2022, Lecture Notes on Data Engineering and Communications Technologies
    • Review of opportunistic network: Assessing past, present, and future

      2021, International Journal of Communication Systems
    View all citing articles on Scopus
    1

    Tel.: +40 745 174 359; fax: +40 318 163 028.

    2

    Tel.: +34 954 48 12 93; fax: +34 954 48 73 73.

    3

    Tel.: +44 151 231 2588; fax: +44 151 207 3460.

    View full text