The concept of ad hoc networking [21–23] is neither new, nor specific to the wireless case.
The basic idea behind ad hoc systems stems from the early stages of Internet development during the cold war, where a distributed network of peer nodes capable of operating even when a number of nodes or links are brought down or destroyed was envisioned. The same idea for distributed operation holds for wireless ad hoc systems, which of course have some additional characteristics that stem from the use of wireless transmission. Thus, the term
‘wireless ad hoc’ stands for a network having no central administration and comprises mobile nodes that use wireless transmission. As seen later, nodes in an ad hoc network can serve as routers as well, by forwarding packets between stations that are out of transmission range of one another. This section aims to introduce the ad hoc concept as this is discussed in detail in Chapter 10. Furthermore, many of the technologies presented in this book, such as the IEEE 802.11 and HIPERLAN WLANs (Chapters 9 and 10) and Bluetooth and HomeRF PANs (Chapter 11) employ ad hoc functionality.
The major characteristics of ad hoc wireless networks are the following:
† Distributed operation. The ad hoc concept differs from other wireless systems, such as cellular systems in terms of network operation. An ad hoc network comprises stations that have the same capabilities and responsibilities. No centralized entity that controls the network exists. In an ad hoc network there are no BSs or MSCs and thus all network protocols operate in a distributed manner.
† Dynamic topology. In a wireless ad hoc network, nodes are free to move in almost any possible manner. The fact that (a) some mobile stations may be out of range of one another and (b) the wireless medium condition changes rapidly over time results in dynamic network topologies with the nature of topological changes being unknown to the network a priori.
† Multihop communications. Due to signal fading and the finite coverage of mobile trans- mitters, a fully connected topology cannot be assumed for an ad hoc system. Thus, in the case where a station A needs to send data to another station B out of its range, the transmission needs to be relayed through other nodes. Such networks are known as multi-
Wireless Communications Principles and Fundamentals 81
hop (or store and forward) wireless ad hoc networks; some examples are the HIPERLAN 1 WLAN standard and Bluetooth.
† Changing link qualities.This is true for all wireless systems, however, it is more important in the multihop case, since the quality of a multihop path depends on the qualities of all the links that make up the path. Thus, monitoring of link quality is bound to be more difficult in the multihop case.
† Dependence on battery life. This applies to most wireless systems, however, in ad hoc systems it is even more important. Consider, for example, the case of cellular systems: BSs are not hindered by finite battery life and overall network performance does not drop when some mobiles switch off due to battery depletion. Rather, it reduces the amount of inter- ference and channel contention and thus increases overall network performance. On the other hand, efficient network operation in ad hoc systems depends on the battery-depen- dent mobile nodes, which are responsible for relaying other nodes’ messages when communicating stations are out of range. A fewer number of nodes results in limited support for relaying and this leads to a network with less routing capability.
In recent years, there has been a big interest in wireless ad hoc networks. This is due to the fact that they possess advantages for certain types of applications, such as emergency systems or military communications, that that need quick deployment of a network in cases where a fixed wireless communication infrastructure does not exist or cannot be used due to security, cost, or safety reasons. Since wireless ad hoc networks can be deployed without needing support for a centralized entity, they are very popular in such situations. Similarly, they are useful in applications where increased network reliability is demanded in cases of failing or departing terminals. An example of this is again military applications where wireless ad hoc networks are very efficient due to the fact that the network does not rely on some critical nodes for its organization or control.
The characteristics of dynamic topology and multihop communications make the design and operation of ad hoc systems a challenging task. Such systems need to operate efficiently even in cases of unknown network topologies and absence of direct paths between commu- nicating stations, which leads to multihop connections. It is evident that the performance of ad hoc systems greatly depends on the efficiency of the routing scheme being used. Thus, wireless ad hoc routing algorithms should be efficient for performing their functions; the most common of these are described in the following subsections.
2.9.1 Network Topology Determination
Ad hoc routing protocols must monitor and react to the changing network topologies. Ad hoc systems may employ multihop communications, thus routing protocols must make sure that at least one path exists from any node to any other node. The only case when this is not demanded is, of course, the case of partitioned networks, where the ad hoc network is split into a number of partitions due to the fact that any two nodes belonging to different partitions are not within range of one another. In order to efficiently monitor and adapt to changing network topologies, ad hoc routing protocols must provide all nodes with knowledge regard- ing their neighbors (those nodes of the network with whom they can directly communicate).
Due to the distributed nature of ad hoc wireless networks, it is obvious that monitoring of network topology will be done in a distributed manner and information regarding the status of routes should be propagated to all network nodes when topology changes occur.
As an example of network topology determination, we present the case of ad hoc network establishment. Figure 2.42 shows an ad hoc network where nodes join the network one after the other, according to the corresponding numbering. Thus, the network can established when node N2comes within the range of node N1. Both these nodes announce their transmission to form a network by regular beacon transmissions that contain information such as their addresses. Assuming that nodes N1and N2establish direct communication, an ad hoc network is formed and the routing protocol updates the routing tables in these nodes so as to reflect the change in topology. When a third node, N3enters the network, the routing tables are updated to reflect the new topology. If N3is within range of both N1and N2, then each node’s routing table contains the possible routes from this node to all others. In this case there obviously exist two routes between each pair of nodes. One is direct and the other is relayed through the third node. When N3is within range of only N1(as shown in the figure) or N2, the routing tables in the nodes are updated correspondingly.
2.9.2 Connectivity Maintenance
After a network’s establishment, topological changes are sure to occur either due to node mobility/failure or changing signal propagation characteristics. Thus, routing protocols need to find alternative routes between stations in order to maintain connections. Consider, for example, the case of the ad hoc network in Figure 2.42. If nodes N3and N1moves so that N3 goes out of range of N1and comes into range of N2the topology changes to that of Figure 2.43. In this case N3and N1can still communicate, although only via node N2. This fact is first detected by N2and N3,which update their routing tables and is then communicated to all the other nodes of the network.
The performance of a wireless ad hoc system greatly depends on the routing protocol’s ability to quickly (a) find loop-free routes between stations when topology is changed and (b) disseminate this information to all the nodes of the network. When network topology changes occur so fast so that the propagation of the previous topology to all nodes update has not yet finished, the performance of the system may degrade significantly. Thus, the application of a specific routing protocol is useful in cases when network topology changes sufficiently
Wireless Communications Principles and Fundamentals 83
Figure 2.42 An ad hoc wireless network
slowly, so as to enable successful propagation of previous topology updates. Wireless ad hoc networks are known as combinatorial stable if and only if they satisfy this constraint.
2.9.3 Packet Routing
As mentioned above, routing schemes are responsible for propagating changes and compute updated routes to a destination when changes in the network topology take place. In order to take into account the characteristics of wireless ad hoc networks, routing protocols for such networks employ a number of additional metrics, apart from the end-to-end throughput and delay metrics that are used in routing protocols for wired systems. The main performance metrics for wireless ad hoc network routing protocols are the following [22]:
† Maximum end-to-end throughput
† Minimum end-to-end delay
† Shortest path between communicating stations
† Minimization of overhead due to control signaling of the routing protocols
† Adaptability to changing topology
† Minimization of total power consumption within the network.
Of course, reaching the optimum values for all of the above constraints cannot be achieved.
Rather, routing protocols provide trade-offs between these metrics. For example, the on- demand routing family of protocols, which is examined in Chapter 10, reduce control over- head at the expense of increasing the time needed to calculate new routing information, thus resulting in increased end-to-end delay.
2.9.4 The Semi Ad Hoc Concept
Another wireless networking concept that is related to ad hoc is the semi ad hoc concept. This concept is possible in cases when many radio access networks are available at the same time.
In such a case devices can implement a dual mode of functionality, thus having the ability to operate either within a wireless network with centralized control (such as a cellular network) or within a wireless ad hoc network. This approach can increase the robustness of wireless
Figure 2.43 Topology change due to mobility
systems. Whenever the entity responsible for the centralized control fails, or users move out of range of the cellular system, devices can set up an ad hoc network of their own. Further- more, when a few network nodes are still in range of the cellular system, their membership in the ad hoc network will provide coverage extension to the other nodes as well. As can be seen in the corresponding chapters, most commercial ad hoc systems such as 802.11 WLANs, Bluetooth and HomeRF can follow this approach, since they can exploit existing wireless infrastructure to expand the set of services offered.
Ad hoc networks are mainly used by the military whereas most commercial systems are centralized. The integration of the various radio access networks into a combined network with seamless mobility, which is envisioned to be achieved by Fourth Generation (4G) wireless networks makes the semi ad hoc concept a promising approach.