Mobile multihop Ad Hoc networks are collections of mobile nodes connected together over wireless medium. These nodes can freely and dynamically self-organize into arbitrary and temporary, "ad hoc" network topologies, allowing people and devices to seamlessly internetwork in areas with no preexisting communication infrastructure.
My research activity (mainly from 2002 to 2006) focused on routing and middleware issues in MANETs based on an experimental approach. In fact, evaluations results have been obtained by the development and experimental testing of routing and middleware protocols in small and medium-scale topologies. All the work has been done in the framework of the EU funded project MobileMAN.
Several routing protocols have been proposed and developed for MANETs, and they can be mainly divided into three categories: proactive, reactive and hybrid. Proactive protocols (based on the Link-state model of routing for Internet) provide a periodical exchange of routing messages in broadcast in order to develop routing tables covering the entire network topology. They mainly guarantee low delays for the network reconfiguration in case of mobility or disconnection events, but they introduce not-neglibible overhead on the network due to the periodical exchange of messages. Instead, reactive protocols are designed to provide a route between a pair of nodes not able to directly communicate only if requested by upper-layer applications. In this way, the network overhead is reduced but each node has only a partial knowledge of the network topology, requiring thus higher delays to support mobility or disconnection events.Hybrid solutions represent a mix of the above mentioned categories. This activity focused on the experimental evaluation of a reactive (AODV) and a proactive (OLSR) routing protocol in small and medium scale testbeds.
Ad hoc networking shares many concepts, such as distribution and cooperation, with the peer-to-peer (P2P) computing model. A defining characteristic of p2p systems is their ability to provide efficient, reliable, and resilient message routing between their constituent nodes by forming virtual ad hoc topologies (i.e., overlays) on top of real network infrastructures. The difference with traditional distributed computing systems is the lack of a central authority that controls the various components. Nodes form a dynamic and self-organizing system. The applications best suited for P2P implementation are those where centralization is not possible, relations are transient, and resources are highly distributed (e.g., File Sharing, distributed search and indexing, resource storage, collaborative work.)
p2p constitutes a natural computing model for ad hoc networks, sharing most of their characteristics. Thus, integrating p2p systems on top of ad hoc networks makes the variety of p2p applications and services available also to mobile users. Specifically, this activity focused on structured overlay networks in which peers organize themselves in a distributed search index that usually contains information on the exact location of each shared data. They define a subject-based routing to distribute and retrieve data among nodes belonging to the same overlay. Each node is identified by a logical address and each data is identified by a pair <Key, Value>. Each data is stored on the node that is logically nearest to the value of the key.
The research focused on the experimental evaluation of an open source implementation of Pastry, a known structured overlay network based on DHT, on real small and medium-scale ad hoc networks. Obtained results showed several drawbacks of applying this solution as-it-is on MANETs mainly due to the overhead introduced to maintain the overlay data structures especially in case of mobility and disconnection events. Thus we proposed a novel, optimized solution of overlay network on MANETs: CrossROAD, Cross-layer Ring Overlay for AD hoc networks. This has been the subject of my PhD thesis.
CrossROAD follows the main principles of Pastry, optimizing them through a cross-layer architecture. This allows protocols belonging to different layers of the legacy network stack to interact through a module that acts as a shared memory. Thus, CrossROAD, which resides at the middleware layer, is designed to interact with a proactive routing protocol in order to access to network topology information. In this way, using IP addresses of nodes as identifiers, CrossROAD is autonomously able to manage the overaly network simply accessing the routing tables and hashing the IP addresses of nodes running its services. In addition, since the proactive routing protocol timely react to mobility and disconnection events, CrossROAD is able to maintain the overlay even in this conditions. More details on CrossROAD design, implementation, and experimental evaluation can be found in my PhD thesis.
Group-Communication Applications are an outstanding opportunity to
bring MANETs in the real world implementing multicast solution to
disseminate and recover contents distributed over the network. These
application are well-supported by structured overlay neworks and
they can be further optmized for MANETs exploiting CrossROAD
features. This activity started with the development of a P2P
Group-Communication Application for MANETs (The Whiteboard) and its
evaluation as a legacy implementation on a multicast protocol
(Scribe) developed on top of Pastry on real testbeds. Then, we have
designed and implemented the same application and its multicast
support on top of CrossROAD, obtaining significant results.
For technical details please see Publications.