P2P over MANETs: Application and Network Layers' Routing Assessment

Introduction

Peer-to-Peer (P2P) networks have emerged as a solution for data sharing as well as processing in distributed environments (Borg, 2003; Talia & Trunfio, 2003; Oliveira, Siqueira, & Loureiro, 2005), and have been rapidly and widely adopted in the Internet.

At the same time, Mobile Ad hoc Networks (MANETs) have enabled a wide range of novel applications in areas such as medicine, including pre-hospital and in-hospital emergency care, rescue team communication in disaster situations, and exchange of information in battle fields (Zhou & Haas, 1999; Borg, 2003), situations where it is not possible to rely on previous infrastructure.

The P2P paradigm (Oram, 2001) is the basis for both P2P networks and MANETs. One of the most significant characteristics of the P2P paradigm is the fact that central units, which manage and meet the needs of the network, are non-existent. In this model, nodes have equivalent features and capabilities and as a consequence, are called “peers”. Every peer thus is able to launch and answer requests originated from any other peer. Additionally, both MANETs and P2P networks are self-organizing networks with dynamic topology responsible for routing queries in a distributed environment.

MANETs and P2P networks are not only similar, but complement each other. Nodes in a MANET typically have low computing capacity and, therefore, are unable to play the role of servers. On the other hand, a P2P application appears to be a powerful tool to disseminate information in MANETs. In other words, since a P2P network does not possess a unique service provider at a certain time, the distribution of tasks among nodes prevents them from becoming overloaded.

Some MANET-enabled applications (e.g. rescue team communication in a disaster scenario and information exchange in battle fields) are bound to have peers cooperating continuously with other peers. For instance, a rescue team player could request the location of his nearest colleague. Although a central server could store such an information, this approach is not only more expensive (as it requires more hops and frequent location updates) but also less resilient: a single point of failure is not desirable in rescue team situations, and servers are the target of attacks in battle fields. In spite of the synergy between P2P systems and MANETs, their integration is not straightforward.

Routing in MANETs was conceived for client-server applications, which usually employ asymmetric flows (i.e., many-to-one). This characteristic contrasts sharply with P2P systems, which have a many-to-many traffic pattern. P2P networks, conversely, have been designed assuming the conventional Internet infrastructure, i.e., a static and fixed underlying topology. As a result, it is crucial that novel solutions, which would take into account the idiosyncrasies of both P2P and ad hoc networks, be devised for P2P systems operating over MANETs (P2P over MANETs).

In this work, we conduct a detailed routing assessment of P2P over MANETs. Specifically, we carried out the following evaluations: (i) AODV (Perkins & Royer, 1999), DSR (Johnson & Maltz, 2001), and DSDV (Perkins & Bhagwat, 1994) routing protocols under a Gnutella P2P network; and (ii) two P2P protocols Gnutella & Chord (Stoica et al., 2003) over MANET with AODV as the network routing protocol.

The remainder of this work is organized as follows. We first explain the synergy between P2P networks and MANETs. Next, we discuss related works on P2P over MANETs. This is followed by a description of the network characterization process. We then evaluate several MANET routing protocols under P2P networks and provide a discussion of the results. Finally, we conclude the chapter.