<<

. 8
( 87 .)



>>

flooding, whereas the senders initiate the flooding in ODMRP.
To avoid the significant delays in route recovery caused by link failures, in [197] the
authors explore the possibility of using a set of precalculated alternate trees. When a link
breaks, another tree that does not includes the failed link is deployed. An alternative ap-
proach to avoiding problems related to tree/mesh maintenance is implemented in the Ex-
plicit Multicasting protocol [198]. This protocol is designed to operate in a stateless man-
ner where no intermediate node needs to maintain multicast forwarding paths.

1.4.4. TCP Issues
TCP was originally designed to work in fixed networks. TCP provides an effective con-
nection-oriented transport control protocol that provides the essential flow control and
congestion control required to ensure reliable packet delivery. Because error rates in
wired network are quite low, TCP uses packet loss as an indication of network congestion,
and deals with this effectively by making corresponding transmission adjustment to its
congestion window. The mobile multihop ad hoc environment brings fresh challenges to
TCP protocol due to its frequent change in network topology, disconnections, variation in
link capability, and high error rate. In a wireless mobile ad hoc network, packet losses are
usually not caused by network congestion, but by the high error rate from wireless medi-
um and frequent disconnections from mobility, resulting in backoff mechanisms being in-
appropriately invoked [36, 99, 101, 103], thus reducing network bandwidth utilization and
increasing the delay for connection restoration [102]. In addition, variation in link capa-
bility could cause asymmetric links and delayed acknowlegment, which can affect con-
gestion window adjustment as well [98“100]. As a result, standard TCP flow control and
congestion control mechanisms do not work well in mobile ad hoc networks.
Besides physical layers issues, a number of studies have shown that MAC layer and
network layer protocols can have a significant impact on TCP performance as well. Since
link-level reliability is provided by the MAC layer, the error control mechanism used by
the MAC layer can adversely affect TCP performance. For example, interaction between
TCP and MAC layer backoff timers can cause severe unfairness and capture conditions
when CSMA and FAMA are used as MAC layer protocols. Well-defined synchronization
is therefore required between the TCP and MAC layer protocols to reduce the effect of
this interference on TCP Performance [98, 104].
The multihop routing nature of MANET also contributes, to a certain degree, to loss of
performance. Measurements using 2 Mbps 802.11 MAC have shown that TCP throughput
27
1.4. TECHNICAL CHALLENGES AND RESEARCH OVERVIEW


decreases by 50% when the traffic moves from the one-hop to the two-hop path [102]. The
study in [36, 102] further shows that when the number of hops is small, the throughput de-
creased with increased number of hops, and was stabilized by effective pipelining only
when the number of hops became large enough.
Finally, different TCP implementations can result in different TCP performance; for
example, conflicts between TCP data packets and TCP ACKs can cause TCP performance
to degrade when window size is greater than 1 packet [104, 105]. Consequently, in order
for the TCP protocol to work properly and effectively in MANET, ad hoc specific adapta-
tions are required at various layers. Numerous enhancements and optimizations have been
proposed over the past few years to improve TCP performance, many of them developed
specifically for wireless cellular networking environments [108“114] where the last hop
is based on a wireless medium. Although there are a number of differences between cellu-
lar and ad hoc networks, many of these proposed solutions can be readily used in the mo-
bile ad hoc networks [98], whereas others can be used after some adaptation. For example,
as packet loss usually results from limitation of the wireless medium, the solution can be
to simply retransmit the lost packet to avoid invocation of congestion control mechanisms.
Besides these techniques, numerous new TCP optimization mechanisms have been pro-
posed with the aim of resolving MANET-specific issues, including the adaptation of TCP
error detection and recovery strategies in ad hoc environments. For instance, methods have
been developed to distinguish between packet losses caused by network congestion/over-
loading and other factors, such as buffer overflow or transmission errors, and mobility by
using link contention information [106], which would allow TCP to take the appropriate ac-
tion. Techniques have been proposed to minimize the impact of mobility and link discon-
nection on TCP performance, such as the use of explicit link failure notification (ELFN)
[102], a technique to detect and respond to out-of-order packet delivery events [103], as
well as link-layer adaptive spacing and link RED methods to adapt TCP for multihop ran-
dom early detection (RED), like graceful drop behavior [106]. To reduce the interference of
the MAC layer on TCP Performance, a new MAC protocol, MACAW, has been proposed to
extend MACA by adding link level ACKs and using a less aggressive backoff policy [104].
A combination of link-level protection, backoff policy, and selective queue scheduling
techniques has been shown critical for efficient and fair operation of ad hoc networks under
TCP [104]. Studies have also been done on comparing different TCP implementations us-
ing metrics such as Throughput, Goodput, TransferTime, Weighted Route Length, Average
Packet Delay [101], and Expected Throughput [102], as well as on effective TCP imple-
mentation techniques, such as how to achieve optimal value for TCP congestion window
size to maximize TCP throughput and reduce packet loss [106].

1.4.5. Energy Conservation
Mobile devices rely on batteries for energy. Battery power is finite and represents one of
the greater constraints in designing algorithms for mobile devices [153“155]. It is there-
fore vital that power utilization be managed efficiently by identifying ways to use less
power, preferably with no impact on the applications. Energy conservation is not restrict-
ed to a single network layer, but instead requires a coordinated effort from all related lay-
ers, including the physical-layer transmissions, the operating system, and the applications
[156]. Research in this area has focused on several aspects, including study of energy con-
sumption behavior at the network interface level [24] in portable wireless devices, and
comparisons of different MAC and routing protocols in terms of their energy conservation
28 MOBILE AD HOC NETWORKING WITH A VIEW OF 4G WIRELESS: IMPERATIVES AND CHALLENGES


capabilities [26, 27]. A first result of designing MAC protocols specifically oriented to re-
duce energy consumption was given in [34]. A treatment of ARQ issues for wireless chan-
nels with the objective of energy minimization was introduced in [196]. Since then, many
other proposals for new energy-aware protocols [30, 32, 33, 35] and energy management
models/techniques [28, 29, 31] have been made.
A sample study investigating the impact of network technologies on power consumption
has been provided in [144]. It has been found that the wireless interface consumes nearly
the same amount of energy in the receive, transmit, and idle states, whereas in the sleep
state, an interface cannot transmit or receive, and its power consumption is highly reduced.
But merely maximizing the time the interface is in power-saving mode (sleep state) is not a
viable approach in an ad hoc network environment, as ad hoc networks rely on cooperative
efforts among participating nodes to deliver the network service. A greedy node that re-
mains most of the time in a sleep state, without contributing to routing and forwarding, will
maximize its battery lifetime but compromise the lifetime of the network.
Strategies have been developed to overcome this problem so that the network interface
can be put in a power-saving mode with a minimum impact on transmit and receive oper-
ations. These policies typically operate at the physical and MAC layers. For example, at
the physical layer, some authors have proposed and analyzed policies (based on monitor-
ing the transmission error rates), that avoid useless transmissions when the channel noise
reduces the probability of a successful transmission [157, 158]. At the MAC layer, energy
conservation can be achieved by reducing the energy required to successfully transmit a
packet, for example by avoiding transmitting when the channel is congested, by synchro-
nizing the node communication time for a single-hop ad hoc network (in 802.11)
[161“163], or by finding intervals during which the network interface does not need to be
listening [160]. For example, while a node transmits a packet, the other nodes within the
same interference and carrier sensing range must remain silent. Therefore, these nodes
can sleep with little or no impact on system behavior.
Other strategies have been developed to achieve energy conservation at the overall net-
work level in addition to the node level strategies mentioned above. For example, when a
region is dense in terms of nodes, only a small number of them need to be turned on in or-
der to forward the traffic so that the overall network lifetime is optimized.
Controlling the power of the transmitting node is the other main method for achieving
power saving in ad hoc networks. Reduced transmission power also allows spatial reuse of
frequencies, which can help increase the total throughput of network and minimize multi-
user interference [86]. In addition, the probability of intercept and detection is lower with
reduced power, which is useful in military applications. On the other hand, reducing trans-
mission power also means a smaller number of feasible links among nodes and, hence,
lower connectivity. These two effects have an opposite impact on energy consumption. A
large part of recent work on energy efficiency in ad hoc networks is concentrated on ener-
gy-efficient routing [164“167], in which the transmitting power level is an additional vari-
able in the routing protocol design [27]. Numerous energy-conscious routing protocols
have been proposed. For example, Minimum Power Routing (MPR) selects the path be-
tween a given source and destination that will require the least amount of total power ex-
pected, while still maintaining an acceptable signal-to-noise ratio at each receiver [89, 91].
It also utilizes physical and link-layer statistics to conserve power, while compensating for
the propagation path loss, shadowing, fading, and interference effects.
Energy-efficiency comparison of a number of MAC-layer protocols [26] examines the
effectiveness of various media acquisition strategies in the presence of contention. Con-
29
1.4. TECHNICAL CHALLENGES AND RESEARCH OVERVIEW


ventional energy-conserving link-layer protocols are designed for centralized environ-
ments where resource-rich base stations are used to control node communication and re-
duce contention via careful scheduling and traffic buffering [24]. In ad hoc environment,
the unpredictable connectivity and limited node buffering capability unfortunately limit
the efficacy of these strategies.
Reference [168] points out battery properties that impact on the design of battery pow-
ered devices. Power-saving policies at the operating-system level include strategies for
CPU scheduling [169, 170] and hard-disk management [171].
It is worth noting that simulation study of the energy consumption of two well-known
ad hoc routing protocols [27] running over IEEE 802.11 demonstrated that an energy-
oriented performance evaluation may lead to quite different conclusions than a band-
width-oriented one when judging protocol performance.
At the application level, conventional strategies used to minimize energy consumption
for wireless nodes are not applicable to ad hoc networks [29, 31]. Policies that exploit the
application semantic or profit by tasks™ remote execution have been proposed [156]. By
utilizing usage patterns associated with user applications such as e-mail and Web brows-
ing, these techniques reduce energy consumption by letting mobile devices spend as much
time as possible in a low-power-consumption sleep state. However, since nodes in ad hoc
networks are involved in forwarding other nodes™ packets as well, it is difficult to predict
the time that a network interface will spend in a low-power sleep state.


1.4.6. Network Security
The wireless and mobile ad hoc nature of MANET brings new security challenges to net-
work design. Because nodes in mobile ad hoc network generally communicate with each
other via open and shared broadcast wireless channels, they are more vulnerable to securi-
ty attacks. In addition, their distributed and infrastructureless nature means that central-
ized security control is hard to implement and the network has to rely on individual secu-
rity solutions from each mobile node. Furthermore, as ad hoc networks are often designed
for specific environments and may have to operate with full availability even in adverse
conditions, security solutions applied in more traditional networks may not be directly
suitable [119, 127].
Understanding the possible form of attacks is the first step toward developing good se-
curity solutions. In mobile ad hoc networks, the broadcasting wireless medium inherently
signifies that an attack may come from any direction and from different layers (network or
application transport such as TCP flooding and SYN flooding). Possible attacks include:

Passive eavesdropping
Denial-of-service attacks
Signaling attacks: Attackers may inject erroneous routing information [122] to di-
vert network traffic, or make routing inefficient
Flow-disruption attacks: Intruders may delay/drop/corrupt all data passing through,
but leave all routing traffic unmodified [121]
Resource depletion attacks: Intruders may send data with the objective of congest-
ing a network or draining batteries [121]
Data integrity attacks, by accessing, modifying, or injecting traffic
Stolen device attacks
30 MOBILE AD HOC NETWORKING WITH A VIEW OF 4G WIRELESS: IMPERATIVES AND CHALLENGES


Several unique solutions have been proposed to address these possible attacks in MANET.
Similar to wireline networks, protecting access to wireless network infrastructure is obvi-
ously the starting step. Many authentication techniques have been proposed to achieve ac-
cess control and data integrity. To prevent attackers from injecting erroneous routing in-
formation and data traffic, use of digital signatures to authenticate a message has been
proposed [120, 122]. Implementation of these schemes requires a certification authority
function to manage the private“public keys and to distribute keys via certificates. Since
certification authority function is not possible in MANET, this function needs to be dis-
tributed over multiple nodes. Reference [120] defines the specific message formats to be
used to carry the digital signature. Other methods for access control include the use of the
resurrecting duckling technique, in which a mobile device will trust the first device that
sends a secret key [128].
Besides authentication, encryption can be used to achieve confidentiality and hide in-
formation during transmission or to store information more safely to prevent passive
eavesdropping and data integrity attacks via using encryption and decryption keys. But
even with encryption, an eavesdropper may be able to identify the traffic pattern in the
network and obtain the mode of operation information. Results in [124] suggest that such
traffic analysis can be prevented by presenting a constant traffic pattern independent of
the underlying operational mode or insertion of dummy traffic.
Other intrusion-related mechanisms proposed include techniques for intrusion-resis-
tant ad hoc routing algorithms (TIARAs) [121] and intrusion detection techniques [123]
that enable early detection of intrusion in the network by using intelligent protocols or
characteristic “training” data such as rate of change of routing information to identify ab-
normal media access patterns, or abnormal routing table updates.

1.4.7. Simulation and Performance Evaluation
Simulation plays an important role in MANET technology development. Constructing a
real ad hoc network test bed for a given scenario is typically expensive and remains limit-
ed in terms of working scenarios, mobility models, etc. Furthermore, measurements are
generally non-repeatable. For these reasons, protocol scalability, sensitivity to user mobil-
ity patterns, and speeds are difficult to evaluate on a real test bed. Using a simulation or
analytic model, on the other hand, permits the study of system behavior by varying all its
parameters and considering a large spectrum of network scenarios. For mobile ad hoc net-
work evaluation, simulation modeling is preferred over analytical modeling due to its
flexibility and ability to model network-level details. A detailed discussion of methods
and techniques for MANETs simulation can be found in [173].
The ability of ad hoc network protocols to correctly behave in a dynamic environment,
where device position may continuously change, is a key issue. Therefore, modeling
users™ movements is an important aspect in ad hoc network simulation. Important aspects
that need to be considered [173] during simulation include the definition of the simulated
area in which users™ movements take place, the rules for modeling users that move beyond
the simulated area, the number of nodes in the simulated area, the node mobility model,
and the allocation of nodes at the simulation start-up, and so on.
Typically, simulation studies consider a fixed number of users that move inside a
closed rectangular area. Rules are defined for users arriving at the edges of the area.
The random waypoint mobility model is the most commonly used technique to define
the way users move in the simulated area. According to this model, nodes move according
31
1.4. TECHNICAL CHALLENGES AND RESEARCH OVERVIEW


to a broken-line pattern, standing at each vertex for a model-defined pause time (p).
Specifically, each node picks a random destination in the rectangular area, samples a
speed value according to a uniform distribution in the range (0, vmax], and then travels to
the destination along a straight line. Once the node arrives at its destination, it pauses for a
time p, then chooses (draws) another destination and continues onward. The pause time
and the maximum speed, v, are mobility parameters. By changing these values, various
system mobility patterns are captured. For example, p = 0 signifies that all nodes are al-
ways in motion throughout the simulation run.
In order to establish a repeatable simulation environment and make for a valid com-
parison of results, a set of network environments and performance metrics have been
proposed. Common environment metrics used to define the networking context include
network size (number of nodes), network density, capacity, connectivity structure (aver-
age number of neighbors, transmission range), mobility pattern (speed, range, direction,
frequency, etc.), link bandwidth (bps), traffic pattern (packet size, transmission frequen-
cy, type of traffic), link characterics (bidirectional or unidirectional), transmission medi-
um (single vs. multichannel), and so on. Commonly used network performance metrics
[96, 97] include network settling time, network join time, network depart time, network
recovery time, route acquisition time, memory required, maximum number of supported
network nodes, frequency of control updates, overhead ratio, number of data packets de-
livered correctly, energy consumption, percentage of out-of-order delivery, end-to-end
data throughput and delay, as well as associated mean, variance, and distribution, and so
on.
A very large number of simulation models have been developed to study ad hoc net-
work architectures and protocols under varying network scenarios (number of nodes, mo-
bility rates, etc.) and constraints (bandwidth and energy, latency, throughput, or associa-
tion stability, etc.). For example, the existence of a large number of routing protocols in
MANET areas makes the routing protocol selection a difficult task. Simulation consider-
ing the above metrics presents an easier and systematic way to construct the network envi-
ronment; to collect and analyze the required performance metrics to make fair protocol

<<

. 8
( 87 .)



>>