1.2.1 Definitions
Giving a precise definition of wireless ad hoc networks is not an easy task, and the literature on the subject doesn’t provide an unified view of the concept. ad hoc essentially means for the occasion. This implies that these wireless networks are created for a specific purpose and disappear with the condition of their creation.
Figure 1.1: Example of multi-hop forwarding in a wireless ad hoc network.
From this common and generic definition, we can deduce the main characteristics of an ad hoc network [107, 136]:
• Because the network is created on the fly, it should not rely on a preexisting fixed and/or wired topology. In that sense, ad hoc networks can be clearly dif-ferentiated from cellular networks, where base-stations are needed. According to [76], the system may however have gateways to and interfaces with a fixed network.
• The preceding point often implies a distributed operation of the network, with no central control, all nodes havinga priori the same role. This characteristic has a huge impact on the protocols design.
• Such a network has a priori no determined topology, so that a node may not be in direct radio connectivity with all other nodes in the network. In this case, nodes should be able to store, forward, and thus route the information from a source to a destination (see figure 1.1). With this capability, nodes are both terminals and routers and hence sometimes designated as terminodes [56]. The wireless ad hoc network is said to be multi-hop.
• An energy-constrained operation may be expected because some nodes may rely on batteries for their energy.
• Literature does not provide a definite answer to know whether or not an ad hoc network is mobile. [76] considers that dynamic topologies are a key char-acteristic of ad hoc networks because nodes are assumed to be mobile.
1.2.2 Some Applications
Ad hoc networks present several advantages. One of the main is to allow a quick deployment, where the telecommunication infrastructure is not available. This can be the case for reasons of cost, safety, security.
The first application of wireless ad hoc networking can be found in the defense environment. In 1972, the american (United States) Defense Advanced Research Projects Agency initiated a research program called Packet Radio Networks [125].
Its aim was to build a packet switched network on the battlefield for mobile users with relaying capabilities. Radio devices were likely to be carried by numerous kinds of supports (aircraft, vehicles, soldiers, ships...) creating dynamic situations. The network had also to be robust to the arrival or the departure of nodes and to link breaks.
After this precursor project, the technology found applications in civilian life in the 1980’s, e.g. the amateur packet radio networks [128]. Today, ad hoc networks are proposed for emergency and rescue operations. After a catastro-phe, an earthquake, or a blast, firemen and rescuers need to communicate with each other and coordinate operations sometimes in total absence of communication infrastructure.
For few years, applications are now anticipated for commercial deployments. We provide here a short list of known projects with examples of applications:
Inter-vehicle communication: Emergency and congestion warnings (figure 1.2), traffic information, lane change assistance, or cooperation at road intersections are provided by multi-hop communications between cars.
Congestion warning
Congestion Area
Figure 1.2: Example of inter-vehicle communication: congestion warning.
Sensor networks: Nodes are sensors in charge of collecting information on a field. As an example, scientific data can be collected by robots on a volcano and transmitted to a center located in a safe zone for analysis.
Home networks: Domestic and electronic equipments (TV, video recorder, fridge, heat controller...) are expected to be wireless enabled and able to communi-cate with each other in a near future via multi-hop networks.
Figure 1.3: Example of meshed network: wireless metropolitan network.
Meshed networks: Backbone inter-connections for wireless networks can be of mesh topology. Such networks are used to inter-connect network equipments like access points (see figure 1.3). Such an inter-connection is foreseen in WiMAX. This can also be a good solution in indoor, where the wired network may not be easily accessible, e.g. in rail stations, hospitals, or harbors.
Figure 1.4: Example of coverage extension.
Coverage extension: The coverage of cellular networks, e.g. WLAN, can be extended thanks to multi-hop networks. The information of a terminal out of the range of an AP can be forwarded by intermediate terminals or specific relay nodes (figure 1.4).
1.2.3 Some Research Areas
Practically all protocol layers have to be specifically tuned to handle the dynamics of wireless transmission, distributed topology and distributed processing inherent in ad hoc networks. This paragraph lists some of the main research interests in the field.
Physical layer: Since ad hoc networks are wireless, research efforts on the transmission over radio channels is a requirement. In the field of multi-hop networks, it can be noted that several papers focus on the cooperation between nodes, and in particular on the performance of the relay channels (see e.g. [81, 109]). The design choices at the physical layer have also a great impact on higher layers. The MAC protocol may for example take into account the underlying modulation, e.g.
Orthogonal Frequency Division Multiplexing (OFDM) [134] or Ultra Wide Band (UWB) [41]. Moreover, ad hoc networks are fundamentally bandwidth-constrained, this can lead to the consideration of high efficiency techniques like multi-packet reception [13].
Synchronization: This is a very critical issue for all distributed TDMA schemes.
A possible solution in outdoor, now at low cost, consists in making use of the GPS (Global Positioning System) that provides a global synchronization for all nodes.
Also the European satellite navigation system, GALILEO, will provide a very good
timing accuracy [85]. In this case, guard intervals have to be foreseen in MAC protocols. Another way of research is local synchronization, where nodes try to syn-chronize themselves by exchanging beacons with their neighborhood [79, 80, 147].
Routing layer: This is probably the most active research area in ad hoc net-works, especially since the establishment of the IETF MANET working group, whose aim is to standardized IP unicast and multicast routing. More than thirty proposi-tions of protocols have been submitted. Four of them are or are expected to become Experimental Request for Comments (RFC) documents: two reactive ones, Dynamic Source Routing (DSR) [122] and Ad hoc On Demand Distance Vector (AODV) [159], and two proactive ones, Optimized Link State Routing (OLSR) [74] and Topology Broadcast based on Reverse Path Forwarding (TBRPF) [156]. Reactive protocols look for routes on demand, while proactive ones continuously maintains routing tables.
Transport layer: Papers on the transport layer have the two fold objective to study TCP over ad hoc networks and to propose modifications or alternatives to it, see e.g. [98, 143, 184].
Energy efficiency: Minimum energy consumption in wireless communication devices is one of the major challenges for designing ad hoc networks. For the obvious sake of cost and portability, the battery-life of a wireless device has to be maximized, while maintaining network connectivity [168]. This issue has been addressed at the MAC layer (see e.g. [175]), and at the routing layer (see e.g. [176] or the power aware modification of Link State and AODV protocols [52])
Node cooperation: In multi-hop networks, a node may need to benefit from the cooperation of other nodes for packet forwarding. Some nodes could however deny this cooperation to save their battery power or even by bad behavior. Such a behavior impacts the network performance negatively. Several algorithm try to tackle this issue (see e.g. [58, 149, 150]).
Higher layers: Security, authentication, authorization, and accounting have specific issues in ad hoc networking [203]. Addressing is of particular interest in networks, where nodes have multiple radio interfaces, and/or can communicate with nodes on a wireline network via gateways (see e.g. [67]). Auto-configuration, service discovery, and address allocation are also active areas of research.
Information theory: In this field, a landmark paper is that of Gupta and Kumar [104] that tries to answer the question: How much information wireless networks can transport? In this paper, interference is treated as noise and nodes are
supposed to be fixed. In this case, it is proved that ifnnodes capable of transmitting W bits/s are in a disk of area 1 m2, the network can transport in the best case Θ(W√
n) bits-meters/s. In random networks, each node can obtain a throughput of Θ(W/√
nlogn) bits/s. [78, 102] have shown that these bounds can be overcome for networks with mobile nodes. These results will be detailed in chapter 4. In [105], Gupta and Kumar have extended their result to cooperative networks: Interference is not seen as noise but may be a source of information. Another recent work [188]
presents a mathematical framework based on rate matrices for finding the capacity region of ad hoc networks.
Following [82], it can be noted that the use of information theory methods and concepts to communication networks is still widely unexplored.
MAC layer: This is historically one of the first issues of research in ad hoc networking [124]. The state of the art in this field is described in the following sections.