<<

. 52
( 87 .)



>>

per, “The most profound technologies are those that disappear. They weave themselves
into the fabric of everyday life until they are indistinguishable from it.” One of those tech-
nologies is computing. Computing will become truly ubiquitous when connectivity is
seamless and its transparency assumed. Self-organizing, adaptive network technologies
like MANET are a necessary component”a step toward the realization of Weiser™s vision.


9.2.2 The Case of Fixed Infrastructureless Operation
Completely fixed infrastructureless operation is another important potential aspect of
MANET technology and can support a wide variety of present and future envisioned ap-
plications. In a completely untethered communication scenario, there may be no connec-
267
9.2 MANET TECHNOLOGY APPLICATION


tion to the greater Internet, but participating users may wish to form a cooperative infra-
structure. Within this infrastructure, cooperative nodes can share data and services, and
potentially perform a host of other tasks. Ad hoc conferencing or business meeting capa-
bilities and ad hoc homeland defense and disaster relief networks are examples or the
more esoteric ad hoc applications. A MANET operating in an autonomous fashion from
the Internet or a greater network infrastructure likely has no fixed access nodes or gate-
way points to provide configuration coordination, so such networks will likely require
more distributed forms of autoconfiguration, service discovery, and management. The
rapidly evolving area of peer-to-peer technologies may be highly appropriate for these
sorts of scenario applications as well, given the likelihood that there will be no presumed
infrastructure hierarchy. Participating devices will likely have to operate in a peer-to-peer
fashion with appropriate applications and protocols, or nodes will have to dynamically
elect and discover distributed services as the network region becomes operational and as
nodes with different capabilities and missions join and leave. Distributed election strate-
gies may also play a role in more robust operations of distributed services where required
in these scenarios.

9.2.3 Other Network Application Areas (Cooperatives and Sensors)
A new area of growing Internet access and deployment interest around the world is in the
area of cooperative, community-based networking. In this case, a community of interest
(e.g., small town government, infrastructure-lacking world region, group of interested in-
dividuals/club) could own and operate a network infrastructure in a cooperative fashion,
much as local roads supplement larger highway infrastructures. If such networks deployed
MANET technology to support a self-organizing, adaptive infrastructure, it would likely
be similar to the hybrid infrastructure extension application we described previously. One
promising example is that of disadvantaged rural regions or developing countries that may
be able to build and operate inexpensive, network infrastructure services with the help of
MANET technology. Such regions often lack the resources or environment suitable for at-
tracting significant fixed-infrastructure developments and services.
More capable and scaled sensor network applications are of growing interest both for
commercial, environmental, and military applications. MANET technology is being con-
sidered as one technology component to support broad applications of self-organizing and
distributed sensor networks. As with generic communication applications, the advent of
small, inexpensive, low-power sensing and computing devices results in novel opportuni-
ties for networked sensor applications. Large-scale deployment of networked sensors for a
variety of applications is becoming more feasible, both technically and economically. As
in our previous examples, both more systematic (e.g., hybrid grid) and ad hoc deploy-
ments of sensor networks are of interest. Not all sensor network applications will desire to
use Internet and MANET technology, but many applications will want to take advantages
of adaptive self-organization and the benefits of building upon the IP design suite. This is
especially true as the role of sensor devices becomes more programmable, adaptive, and
cooperative in a network setting. Developments in this application area are likely to con-
tinue to grow in interest and maturity over the coming years.

9.2.4 Large-Scale versus Small-Scale Use
MANET technology has been envisioned for deployment in many different application
scenarios. We have discussed a few broad related application areas in this chapter. A clear
268 ROUTING TECHNOLOGY FOR DYNAMIC WIRELESS NETWORKING


distinction needs to be made between simple, self-organizing networks operating on small
scales (standalone, moderate-size stub networks, or limited hybrid extension grids), and
those expected to perform adequately in regions with a large set of peer routing nodes
with every node acting as an active router. Large-scale ad hoc networks face a myriad of
difficulties. Conceptually, the core problems faced”sensing and adjusting to dynamic
conditions”are the same on small and large scales. But in practice, networked communi-
cation dynamics favor smaller wireless networks or at least networks requiring fewer hops
in order to access a fixed backbone. For example, in ad hoc networks employing omnidi-
rectional antennas, recent research [6] has shown that as the network scales (assuming it is
equally likely that any pair of network nodes wish to communicate) as the number of com-
municating network nodes grows, the achievable throughput per node pair goes to zero.
This analysis does not even consider the bandwidth required to run a routing protocol. As
such, the prospect of deploying very large-scale ad hoc networks based on broadcast
transmissions is not very promising if the level of nonlocal communication is nonnegligi-
ble.
More promising, perhaps, in large-scale deployment of MANETs is operating in a hy-
brid fixed/ad hoc grid fashion, on the edge of a fixed, high-capacity network. In such de-
ployments, traffic destined to topologically proximate MANET nodes would be forward-
ed directly in the MANET cloud, whereas traffic destined for topologically distant
MANET nodes would be forwarded to a MANET/fixed network gateway near the sender,
then through the fixed network to a similar gateway near the destination, and, finally,
through the MANET cloud to the destination. There are many ways to envision realizing
such hybrid fixed/MANET interoperability and we discussed some of the routing aspects
of a scenario depicted in Figure 9.7.
One possibility is for IPv4 MANET nodes to use as their host-routed interface address
in the MANET network an address that is also a Mobile IP (MIP) Home Address (HoA)
registered in a fixed network MIP Home Agent (HA) with reachability to all relevant
MANET gateways. If a given IP address does not appear in a MANET node™s routing
table, its default forwarding behavior is toward the nearest MANET gateway (likely via
some form of MANET anycast routing). Such packets arriving at the gateway are for-
warded to the HA, and then tunneled to the gateway near the MANET destination node.
This MANET gateway is also a Mobile IP Foreign Agent (FA) that advertises its Care-of
Address (CoA) locally within the MANET via topologically scoped flooding using a
MANET broadcast service. The MANET destination [also a Mobile IP Mobile Node
(MN)], having learned this CoA information, would have registered its HoA-CoA binding
with its HA via the local MANET Gateway/MIP Foreign Agent. Upon reception of the
tunneled packet, the FA will decapsulate the packet and natively route (via MANET rout-
ing) the packet to the MANET destination node. In this fashion, MANETs of literally any
size may be deployed on the edge of a fixed network infrastructure.
Many other interoperability techniques can be and have been proposed as well, includ-
ing approaches for IPv6. The key consideration is that packets be routed off the MANET
cloud as soon as possible. This is because each MANET hop is expensive from a commu-
nication perspective relative to fixed network bandwidth. In fact, due to the numerous
complications present in mobile wireless networks, it is generally desirable to use the
fixed network bandwidth whenever possible. MANET technology, generally presenting
the most stressing wireless networking environment, should typically be applied where
and when the convenience and dynamic networking opportunities afforded by MANET
technology is worth the potential likelihood of reduced performance.
269
9.3 MANET AND RELATED IETF WORK


9.3 MANET AND RELATED IETF WORK

The Internet Engineering Task Force (IETF) is a large, open international community of
network designers, operators, vendors, and researchers concerned with the evolution of
the Internet architecture and the smooth operation of the Internet. The principal products
of the IETF are Internet Standards, generally published in the format of Request for Com-
ments (RFC) documents. The actual technical engineering work of the IETF is largely
done in its working groups, which are organized by topic into several areas.
Although there are many past and present technical efforts dealing with some aspect of
network mobility, an IETF effort that has considered wireless mobility from its inception
in the late 1990s has been the MANET Working Group (WG). The MANET WG™s prima-
ry purpose is to focus on dynamic, wireless IP routing technology and develop and evolve
MANET-related routing specifications and introduce them to the Internet Standards
process. From its inception, the WG has been acutely aware that network operation in dy-
namic wireless environments challenges present requirements and design assumptions of
Internet protocols and applications”requirements and assumptions derived in the past
from a largely wired and tethered network mindset. That is not to say that applications
should be designed specifically for use in wireless networks. Rather, applications and pro-
tocols should be designed to gracefully accommodate changes and degradations in con-
nectivity, changes that may occur more frequently in wireless networking contexts, but
which occur in fixed networks as well due to congestive dynamics and other factors. Prior
to its current status as a topic of Internet standards work, MANET technology had a long
history.


9.3.1 History and Motivation
The technology of MANET is somewhat synonymous with Mobile Packet Radio Net-
working (a term coined during early military research in the 1970s and 1980s), Mobile
Mesh Networking [2] (a term that appeared in an article in The Economist regarding the
structure of future military networks) and Mobile, Multihop, Wireless Networking (per-
haps the most accurate term, although a bit cumbersome). Since its inception at the 39th
IETF in Munich in August 1997, the MANET WG functioned in a part research, part en-
gineering mode, and has developed and fostered a significant number of proposed routing
protocol and analysis methods. The WG has continued to evolve and recently is planning
to convert its scope from the pseudoresearch mode of its origin into a pure engineering
group, with more mature technical work proceeding through group consensus and scoped
near-term problem areas. In parallel with this rechartering, an Internet Research Task
Force (IRTF) MANET Research Group was created as a subgroup of the IRTF Routing
Research Group. The research theme of this research group is to concentrate on more
complex scalability issues and concerns. Several IETF MANET Internet Drafts, and two
basic reactive and two basic proactive protocol designs have reached a reasonable level of
maturity, analysis, and implementation experience. These include the following:

Ad Hoc On-Demand Distance Vector (AODV) [18]
Distributed Source Routing (DSR) [17]
Optimized Link State Routing (OLSR) [16]
Topology Dissemination-Based Reverse Path Forwarding (TBRPF) [15]
270 ROUTING TECHNOLOGY FOR DYNAMIC WIRELESS NETWORKING


Work on the design of the protocols AODV DSR, OLSR, and TBRPF is roughly complet-
,
ed and these protocols will likely be soon considered for Experimental RFC status to en-
courage additional third-party (e.g., industry) experimentation with concepts and imple-
mentations. Lessons learned from the past development of these core protocols will form
an engineering basis for any follow-on efforts of more consensus-based protocol work in
the future. The MANET IETF WG has produced a significant body of prototype MANET
protocols, but work on developing standards will continue to evolve as more is learned
from actual application experience and appropriate Internet engineering efforts in the
coming years.


9.3.2 Some Future Work Issues
Those familiar with wireless network design often realize that layered design is more
complicated than in wired networks due to more severe functional and performance de-
pendencies between the layers. Thus far, the Internet has been designed, and its protocols
optimized, largely for fixed-network deployment scenarios. Mobility is still largely a sec-
ond-class citizen in the Internet standards, and its support has been mostly an after-
thought. Even “Mobile” IP was originally intended to support portability; i.e., the ability
to connect on a foreign subnet using a local address (a colocated care-of address) as a
source address, and yet remain reachable via a second “home” address, possibly stored in
the dynamic name system. Larger system aspects as they relate to mobility, such as hand-
off, were seemingly not originally considered.
The dominant wired network design tradition translates directly to the types of inter-
faces that IP expects to support, and the dynamics assumed to be associated with those in-
terfaces. Layer 2 interfaces fundamentally operate as either broadcast or point-to-point.
From these primitives, additional Layer 3 interface constructs such as nonbroadcast multi-
ple access and point-to-multipoint are created as necessary. This approach has served the
wired Internet well. However a third type of Layer 2 interface is necessary to efficiently
and seamlessly extend IP over dynamic networks, principally wireless ones. This inter-
face, here termed a “dynamic” interface, combines traditional broadcast interface ad-
dressing semantics (i.e., support for unicast, multicast, and broadcast link-layer addresses)
with Layer 2 association event support for the dynamic creation of peer-to-peer interface
associations within an otherwise broadcast interface. Its intended domain of applicability
covers cellular, WLAN, MANET, and so on; in short. all currently envisioned forms of
dynamic wireless networking. The support for this form of interface in all IP stacks would
enable Layer 2 designers to craft link layers that transfer standardized signals to an IP
stack. Simple event notifications such as link active/inactive would facilitate efficient
IP/link layer interaction, without the need for expensive, periodic IP-level signaling (e.g.,
Hello Beacons) to ascertain IP-layer neighbor information. Moreover, with such function-
ality standardized in IP™s underbelly, future link layers may be developed with IP-aware-
ness such that the link active event could immediately convey a neighboring IP address to
an adjacent stack. The IP stack can choose to do what it wishes with such information, in-
cluding ignoring it if it is so configured.
As the understanding of network and wireless dynamics increases, quantitative infor-
mation such as link quality can also be normalized into metrics suitable for consumption
by a dynamic interface-aware IP stack, and conveyed to a routing protocol. It is an open
question as to how much integration of information sharing should occur between a net-
work and lower layers. But beginning even basic work in this direction supports a move
271
9.4 CONCLUSION


toward more optimal design strategies, and good solutions can only improve MANET per-
formance over the present state of the technology.
Context-aware routing is a promising area for future MANET research. The idea is that
the network routing protocol itself can be polymorphic, and be able to adapt its operation
to a changing network membership and deployment context in real time. Realize that the
network itself may be moving and its composition changing. The most appropriate routing
policy may change during the lifetime of a network or, even more interestingly, certain
contiguous regions of a MANET may choose to run different routing policies at the same
time, with MANET routers along the border of different regions serving as dynamic rout-
ing protocol gateways.
Providing Quality of Service other than best-effort or simple Differentiated Services is
a very difficult problem in MANETs. Recalling the previous discussion on integrated de-
sign, much of what is achievable in terms of QoS depends in large part on the link-layer
technology in use. The extent that the particulars of any given link layer can be effectively
conveyed to an IP standard routing protocol in a standardized yet useful fashion remains
to be seen. This is a challenging area of future research. Finally, perhaps the biggest chal-
lenge facing MANET QoS is that many of the link layers responsible for its popularity run
in the unlicensed spectrum. It is simply infeasible to provide strong QoS guarantees in a
spectrum you do not control.


9.4 CONCLUSION

Although significant packet radio work was begun in the 1970s, recent growing interest
and associated technology revolution of MANET wireless technology has fundamentally
resulted from two recent technological events:

The everywhere in everything Internet Protocol (IP) revolution
The embedded, inexpensive, unlicensed wireless technology revolution

Dynamic, wireless IP-based routing has reached a level of practicality and maturity for a
wide variety of scenarios. We discussed four practical, near-term applications for
MANET technology that we feel provide needed capability enhancements for today™s
evolving wireless network applications:

1. Small-to-moderate-sized Internet stub wireless routing regions (see Figure 9.4)
2. Small-to-moderate-sized autoconfiguring wireless networks (see Figure 9.5)
3. Hybrid MANET-enabled access grid regions, handling dynamics yet providing effi-
cient fixed Internet backbone access (see Figure 9.7)
4. Untethered, ad hoc collaborative networks

We briefly discussed related MANET work that has recently been accomplished within
the Internet Standards forum and the potential future direction for MANET work in sever-
al key areas. The understanding gained from these early activities will now be used to de-
velop standardized MANET routing functionality for the Internet. More general lessons
learned in terms of improved robustness and efficiency may also help influence and im-
prove future designs of wired protocols. Despite recent development progress and confi-
dence in some deployment scenarios, there still remains much to do in terms of under-
272 ROUTING TECHNOLOGY FOR DYNAMIC WIRELESS NETWORKING

<<

. 52
( 87 .)



>>