Thus far, the book has discussed mobility in the context of a structured and "centralized" network. While the Mobile Node is away from home, it registers back to an infrastructure or central location, namely, the Home Network, with its Home Agent. Moreover, it can use the infrastructure on the foreign domain through a FA or access router. Thus, not only does the Mobile Node have the freedom to roam, but it also has the luxury of a structured network for support.
Now imagine a mobile environment in which a central infrastructure does not exist. The Mobile Nodes would have to determine and manage their own routing as they and their peers move. This is quite a nontrivial task and has been the focus of much research over the years. The term for such a network is a Mobile Ad Hoc Network (MANET). You might already be familiar with such networks under the term packet radio networks, which have been around since the 1960s in a limited military capacity. The terms are synonymous. With the boom of mobile devices, MANETs are finding broader scope in commercial and military environments. And, as wireless technology advances, users desire even more freedom as they roam.
The aim of these networks is to support robust and efficient communication in a mobile wireless decentralized network by incorporating routing functionality into Mobile Nodes. An MANET has characteristics that make it fundamentally different from most wireline and centralized wireless networks.
An MANET is an autonomous set of nodes distributed over a geographical area that communicate over bandwidth-constrained wireless links. This sounds like a loaded sentence, but basically it means a group of Mobile Nodes that want to communicate with no existing infrastructure. This could be an army battalion, a mobile sensor network, participants in a conference, or even mobile users at a rock concert. Each node in an MANET can represent a transmitter, receiver, or relay station with varying physical capabilities. Typically, a discriminating and valuable capability is the power source because many nodes can operate on a limited battery supply. In such a network, packets usually traverse several intermediate (relay) nodes before reaching their destinations. These networks typically lack infrastructure: Nodes are mobile, no central hub or controller exists, and thus no fixed network topology exists. Although this can place more responsibility on the nodes, it allows mobile networks to be deployed without a priori planning or structure in place.
MANETs must also contend with a difficult and variable communication environment. Packet transmissions are plagued by the usual problems of radio communication, which include propagation path loss, signal multipath and fading, and thermal noise. These effects vary with terminal movement, which also induces Doppler spreading in the frequency of the transmitted signal. Finally, transmissions from neighboring terminals, known as multiaccess interference, hostile jammers, and impulsive interference (for example, ignition systems, generators, and other nonsimilar in-band communications) can contribute additional interference. Sounds a bit chaotic and messy!
Given this nature of MANETs, whether two nodes can directly communicate with one another is a function of their variable link quality, including signal strength and bandwidth. Thus, end-to-end routing paths vary based on environment and resulting network topology. In such networks, the topology can be stable for periods of time and then suddenly become unpredictable. Because MANETs are typically decentralized systems, no central controllers or specially designated routers exist to facilitate routing as the topology changes. All the routing decisions and forwarding (relaying) of packets must be done by the nodes themselves, and communication is on a peer-to-peer basis.
Mobile ad-hoc networking covers a broad spectrum of networking scenarios, ranging from a tactical battlefield to personal peer-to-peer communications. As you could guess, numerous variants of MANETs exist, depending on network size, network density, nodal mobility, and communication environment. The optimal routing approach depends on these various factors, and thus, a notion of the best MANET routing protocol does not exist. For example, an MANET network with high nodal mobility in a hostile environment has different routing requirements than an MANET network with a low rate of topological change.
Regardless of the network scenario, we can draw a few conclusions on the behavior of an MANET. Figure 9-13 shows a three-dimensional plot that characterizes MANET routing. Essentially, this plot shows that the "amount of data" successfully transmitted in the network, the "number of nodes" in the network, and the "frequency of movement" of the nodes are all intimately related. If you think about it for a minute, it is all intuitive. For example, if a network supports many nodes that are moving around often, the amount of data that can be transferred is far less than a network supporting few nodes that are relatively stationary.
Figure 9-13. State of the Art for MANET Routing Protocols
Any routing protocol deployed in an MANET should be able to perform as the network characteristics change, keeping the inferences of the previously described plot in mind. You already know that no notion of a best MANET routing protocol exists because it is not possible to be optimal in all scenarios. The MANET Working Group in the IETF has investigated this problem for several years. During this time, numerous MANET routing protocols have been proposed in the MANET Working Group. In a broad sense, the protocols can be categorized into two groups: reactive (on-demand) protocols and proactive protocols. Proactive protocols incur latency initially when determining all routes, even though they might never be used. Conversely, reactive protocols incur latency during the route discovery phase when a route is needed and also when using an obsolete route in a routing cache. After much technical debate and performance evaluations in the research community, three MANET protocols, as follows, were moved to experimental RFC status, with one on the way:
Not to belabor the point, but no "winner" exists among these protocols. In fact, the performance of a protocol highly depends on the network environment and traffic pattern. For example, proactive protocols perform better in networks with higher mobility, because the protocol overhead to keep routes alive as the topology changes is warranted. On-demand protocols perform better in relatively stable networks with set source-destination communication pairs. Although several research implementations of the protocols can exist, the protocols are still considered to be in an infancy stage. No real-world deployments of the protocols exist, and performance evaluations have been based on testing studies.
Cisco Engineering is taking its own approach by leveraging existing IOS routing protocols, with initial efforts focusing on the Open Shortest Path First (OSPF) protocol. Besides the fact that no winning MANET routing protocol exists, there are several other advantages for this approach, including easier integration of MANETs with existing networks. This would allow mobility devices to more easily migrate between an MANET and an infrastructure network. At the time of this writing, a special design team has been tasked in the IETF to standardize an OSPF-MANET routing protocol. (Cisco is a key participant in the design team.)
Besides routing within the MANET, it is necessary to route across the MANET. An MANET can be a stand-alone network in which the nodes communicate just among themselves. This can be the case in a sensor network, for example. Or, the MANET network might need to communicate across the global Internet. For such cases, integration between the MANET routing protocol and Mobile IP seems imminent. The two can already coexist, with the MANET protocol providing the backend peer-to-peer communications and Mobile IP providing global communication through a gateway node on the MANET registering back to a Home Agent. However, because direct link-layer connectivity between the gateway(s) in the MANET and a FA on the visiting network is likely not always possible, modifications are necessary for viable integration.
Needless to say, the future of MANET routing is still in the works, and research continues. Stay tuned for coming attractions!