Ad Hoc Mobile Networking

 
Ad Hoc Mobile Networking
Several challenges have to be met for an ad hoc network to be possible in a B3G mobile networking
environment. However, most of the challenges that apply to B3G systems in general also apply to
B3G ad hoc networking: spectrum allocation issues, the integration of WMAN/WLAN and cellular
networks, the need for high-speed data in a heterogeneous environment, and the issue of reconfigurability.
The concept of an ad hoc B3G network is somewhat hazy – if a B3G device can always
connect to the seamless, ubiquitous, global network, when would the need for an ad hoc network
ever arise? The answer is, in response to an extraordinary event. The first responders to accidents
may use ad hoc networks for secure and congestion-free communications around the scene. Military
uses for ad hoc networks are well defined. Spectrum allocation takes care of emergency and military
frequency issues. Anyone else who might desire privacy from the ubiquitous, global net (for instance,
for teleconferencing) could establish a private MANET – if they could gain access to some unused
frequency band and keep it “private” for a period of time. Regulators will need to address this issue
for B3G MANETs to thrive.
In general, devices intended for use in a B3G environment in general should be able to scan in
a specific environment to discover candidate available for access networks and register some policy
issues. Devices intended for some use in a B3G MANET should be able to scan for and identify the
frequency band(s) available for temporary, “private” use. In the seamless, global B3G network, the
authentication and authorization mechanisms for access to different networks could be connected to
allow a user/device to move between different access networks without the need to log on multiple
times. In a B3G MANET, the trick would be to keep a device from inadvertently leaving the MANET
and joining the global one. A worse case would involve a MANET node authenticating an undesired
device to the MANET (the undesired device would be scanning for its best connection at all times)
[543]. B3G ad hoc networks should be able to robustly adapt to changing network conditions and
topologies, having the capability to grow, fragment, and reorganize in the absence of centralized,
hierarchical infrastructures [544].
Some issues for B3G MANETs have been identified:
Routing: for different ad hoc scenarios, the routing protocol differs dramatically. While the routing
protocol for an “eHome” scenario can assume fixed wireless terminals (leading to a small dynamic
for the routing), the terminals in a fire-fighting scenario are highly mobile (leading to a high dynamic
for the routing). This leads to the assumption that different routing strategies have to be applied.
Auto-Configuration: If we focus on Internet Protocol (IP) services over ad hoc networks, we have
to support the assignment of IP addresses. Protocols such as dynamic host configuration protocol
(DHCP) will not work in an ad hoc environment.
Device Classes: The routing process depends on the device class of a wireless terminal. Device
classes are based on power, range, air interface, costs, and so on. Terminals with batteries are not
well suited for multi-hop routing since they tend to consume more resources [545].
Table 6.3 shows the characteristics of a variety of MANET technologies.
The specific characteristics of MANETs impose many challenges on network protocol designs on
all layers of the protocol stack. The physical layer must deal with rapid changes in link characteristics.  The MAC layer needs to allow fair channel access, minimize packet collisions, and deal with hidden
and exposed terminals. At the network layer, nodes need to cooperate to calculate paths. The transport
layer must be capable of handling packet loss and delay characteristics that are very different from
wired networks. Applications should be able to handle possible disconnections and reconnections.
Furthermore, all network protocol developments need to integrate smoothly with traditional networks
and take into account possible security problems. The technological challenges that B3G ad hoc network
protocol designers and network developers are faced with include routing, service and resource
discovery, Internet connectivity, billing, and security.
As MANETs are characterized by a multi-hop network topology that can change frequently
because of mobility, efficient routing protocols are needed to establish communication paths between
nodes, without causing excessive control traffic overhead or computational burden on the powerconstrained
devices. Combinations of proactive and reactive protocols, where nearby routes (for
example, maximum two hops) are kept up to date proactively, while faraway routes are set up
reactively, are possible and fall in the category of hybrid routing protocols. A completely different
approach is taken by the location-based routing protocols, where packet forwarding is based on the
location of a node’s communication partner. Location information services provide nodes with the location of the others, so packets can be forwarded in the direction of the destination. Simulation
studies have revealed that the performance of routing protocols in terms of throughput, packet loss,
delay, and control overhead strongly depends on the network conditions such as traffic load, mobility,
density and, the number of nodes. Ongoing research is investigating the possibility of developing
protocols capable of dynamically adapting to the network.
MANET nodes may have little or no knowledge about the capabilities of, or services offered by,
each other. Therefore, service and resource discovery mechanisms, which allow devices to automatically
locate network services and advertise their own capabilities to the rest of the network, are an
important aspect of self-configurable networks. The possible services or resources include storage,
access to databases or files, printers, computing power, and Internet access. “Directoryless” service
and resource discovery mechanisms, in which nodes reactively request services when needed and/or
nodes proactively announce their services to others, seem an attractive approach for infrastructureless
networks. The alternative scheme is directory-based and involves directory agents where services are
registered and service requests are handled. This implies that this functionality should be statically or
dynamically assigned to a subset of the nodes and kept up to date. Existing directory-based services
and resource discovery mechanisms are unable to deal with the dynamics in ad hoc networks. Currently,
no mature solution exists, but it is clear that the design of these protocols should be done in
close cooperation with the routing protocols and should include context awareness (location, neighborhood,
user profile) to improve performance. Also, when ad hoc networks are connected to a fixed
infrastructure (for example, the Internet or a cellular network), protocols and methods are needed
to inject the available external services offered by the service and content providers into the ad hoc
network.
To enable communication between nodes within the ad hoc network, each node needs an address.
In stand-alone ad hoc networks, the use of IP addresses is not obligatory, as unique MAC addresses
could be used to address nodes. However, all the current applications are based on transmission control
protocol (TCP)/IP or user datagram protocol (UDP)/IP. In addition, as B3G mobile ad hoc networks
will interact with IP-based networks and will run applications that use existing IPs such as TCP
and UDP, the use of IP addresses is inevitable. Unfortunately, an internal address organization with
prefixes and ranges as in the fixed Internet is hard to maintain in mobile ad hoc networks owing to
node mobility and overhead reasons, and other solutions for address assignment are thus needed. One
solution is based on the assumption (and restriction) that all MANET nodes already have a static,
globally unique and preassigned IPv4 or IPv6 address. This solves the whole issue of assigning
addresses, but introduces new problems when cooperating with fixed networks. Connections coming
from and going to the fixed network can be handled using mobile IP, where the preassigned IP address
serves as the mobile node’s home address. All traffic sent to this IP address will arrive at the node’s
home agent. When the node in the ad hoc network advertises to its home agent the IP address of the
Internet gateway as its care of address, the home agent can tunnel all traffic to the ad hoc network on
which it is delivered to the mobile node using an ad hoc routing protocol. For outgoing connections,
the mobile node has to route traffic to an Internet gateway, and for internal traffic an ad hoc routing
protocol can be used. The main problem with this approach is that a MANET node needs an efficient
way of figuring out if a certain address is present in the MANET or if it is necessary to use an
Internet gateway, without flooding the entire network. Another solution is the assignment of random,
internally unique addresses. This can be obtained by having each node pick a more or less random
address from a very large address space, followed by duplicate address detection (DAD) techniques in
order to impose address uniqueness within the MANET. Strong DAD techniques will always detect
duplicates, but are difficult to scale in large networks. Weak DAD approaches can tolerate duplicates
as long as they do not interfere with each other, that is, if packets always arrive at the intended
destination. If interconnection to the Internet is desirable, outgoing connections could be realized
using network address translation (NAT), but incoming connections still remain a problem if random,
not globally routable, addresses are used. Also, the use of NAT remains problematic when multiple
Internet gateways are present. If a MANET node switches to another gateway, a new IP address is used and ongoing TCP connections will break. Another possible approach is the assignment of unique
addresses that all lie within one subnet (comparable to the addresses assigned by a DHCP server).
When attached to the Internet, the ad hoc network can be seen as a separate routable subnet – probably
the norm in a B3G environment. This simplifies the decision of whether a node is inside or outside
the ad hoc network. However, no efficient solutions exist for choosing dynamically an appropriate,
externally routable, and unique network prefix (for example, special MANET prefixes assigned to
Internet gateways), handling the merging or splitting of ad hoc networks or handling multiple points
of attachment to the Internet. It is clear that, although many solutions are being investigated, no
common adopted solution for addressing and Internet connectivity is available yet. New approaches
using host identities, where the role of IP is limited to routing and not addressing, combined with
dynamic name spaces, could offer a potential solution, but may be problematic in a B3G environment.
The wireless mobile ad hoc nature of MANETs brings new security challenges to network design.
Because the wireless medium is vulnerable to eavesdropping and ad hoc network functionality is
established through node cooperation, mobile ad hoc networks are intrinsically exposed to numerous
security attacks. During passive attacks, an attacker just listens to the channel in order to discover
valuable information. This type of attack is usually impossible to detect, as it does not produce any
new traffic in the network. On the other hand, during active attacks, an attacker actively participates
in disrupting the normal operation of the network. This type of attack involves deletion, modification,
replication, redirection, and fabrication of protocol control packets or data packets. Securing ad hoc
networks against malicious attacks is difficult to achieve. Preventive mechanisms include authentication
of message sources, data integrity, and protection of message sequencing, and are typically
based on key-based cryptography.
Incorporating cryptographic mechanisms is challenging, as there is no centralized key distribution
center or trusted certification authority at present. These preventative mechanisms need to be sustained
by detection techniques that can discover attempts to penetrate or attack the network. Moreover,
not all security problems in ad hoc networks can be attributed to malicious nodes that intentionally
damage or compromise network functionality. Selfish nodes, which use the network but do not cooperate
in routing or packet forwarding for others in order to conserve battery life or retain network
bandwidth, constitute an important problem as network functioning entirely relies on the cooperation
between nodes and their contribution to basic network functions. To deal with these problems, the
self-organizing network concept must be based on an incentive for users to collaborate, thereby avoiding
selfish behavior. Existing solutions aim at detecting and isolating selfish nodes using watchdog
mechanisms, which identify misbehaving nodes, and reputation systems, which allow nodes to isolate
selfish nodes. Another promising approach is the introduction of a billing system into the network
based on economical models to enforce cooperation. Using virtual currencies or micropayments, nodes
pay for using other nodes’ forwarding capabilities or services and are remunerated for making theirs
available. This approach certainly has potential in scenarios in which a part of the ad hoc network
and services is deployed by companies or service providers (for example, location- or context-aware
services, a sports stadium, or a taxi cab network). Also, when ad hoc networks are interconnected to
fixed infrastructures by gateway nodes, which are billed by a telecom operator, billing mechanisms
are needed to remunerate these nodes for making these services available. Questions about who is
billing whom, and for what, need to be answered and may lead to complex business models. Further
research into security mechanisms, mechanisms to enforce cooperation between nodes, and billing
methods are needed for B3G MANETs to function [546].