Computer Science Technical Report Archive

USC Computer Science Technical Reports, no. 775 (2002)

(USC DC Other)

USC Computer Science Technical Reports, no. 775 (2002)

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1
BRICS: A Building-block approach for analyzing RoutIng
protocols in ad hoc networks - a Case Study of reactive
routing protocols
Fan Bai, Narayanan Sadagopan, Ahmed Helmy
Abstract—One of the main challenges in ad hoc networks
research is understanding the effect of mobility on the per-
formance of routing protocols. In a previous study, we have
shown Why mobility affects the performance of routing pro-
tocols. In this study, we further extend our approach to an-
alyze the interplay between mobility and the protocol mech-
anistic building blocks. Through this approach we hope to
explain How performance varies with mobility by decom-
posing the protocol into parameterized mechanistic build-
ing blocks based on their functionalities. Then, we apply
this approach to reactive MANET routing protocols like
AODV and DSR, which enables us to build a common build-
ing block architecture that encompasses these reactive pro-
tocols. The effect of mobility on each building block is eval-
uated. We are specifically interested in understanding the
contribution of each building block to the overall protocol
performance. Through simulations, several lessons on pro-
tocol design are learnt. For example, in both AODV and
DSR, flooding and caching seem to have a great effect on per-
formance, while salvaging in DSR barely seems to have an
effect on the protocol performance.
I. INTRODUCTION
The emerging field of ad hoc networks has at-
tracted a lot of research attention. This field pro-
vides several challenging problems that differentiate
it from the traditional wired Internet. One main chal-
lenge is understanding the effects of mobility on the
performance of ad hoc routing protocols. Although
it is common practice to evaluate such protocols us-
ing the random way point mobility model, we believe
this is only a first step.
In a previous study, we proposed a framework
called IMPORTANT to systematically evaluate the
impact of mobility on the performance of MANET
routing protocols [1]. The objective of this frame-
work was to answer the following questions:
Department of Electrical Engineering, University of Southern Cali-
fornia, E-mail: fbai@usc.edu
Department of Computer Science, University of Southern California,
E-mail: narayans@cs.usc.edu
Department of Electrical Engineering, University of Southern Cali-
fornia, E-mail: helmy@ceng.usc.edu
1) Whether mobility affects the routing protocol
performance?
2) If mobility affects the performance, Why?
3) If mobility affects the performance, How?
To answer the Whether, we evaluated the MANET
routing protocols like DSR [2], AODV [4] and
DSDV [3] across different mobility models. We used
four mobility models i.e. Random Waypoint(RW),
Reference Point Group Mobility(RPGM), Free-
way(FW) and Manhattan(MH) models. We observed
that different protocols perform differently under the
same mobility pattern. Moreover, the same routing
protocol performs differently under different mobil-
ity patterns as well. The above observations indicate
that mobility pattern does affect the performance
of MANET routing protocols. To answer the Why,
several protocol-independent metrics including mo-
bility and graph connectivity metrics were proposed
to capture the characteristics of mobility and con-
nectivity respectively. From our study, we proposed
the connectivity graph properties as a ”bridge” to
explain Why mobility affects the performance of
routing protocols. A brief description of the mobility
models and the metrics used is given in section II.
In this study, we further extend our approach to
analyze the interplay between mobility and the pro-
tocol mechanistic building blocks. Through this ap-
proach we hope to explain How performance varies
with mobility. In BRICS, the overall protocol is de-
composed into its constituent mechanistic building
blocks. The effect of different mobility characteris-
tics on each building block is then evaluated. We
are specifically interested in understanding the con-
tribution of each building block to the overall proto-
col performance. Such understanding enables us to
reason about the performance difference between dif-
ferent protocols. As a first attempt, we apply our ap-
proach to reactive ad hoc routing protocols; namely,
DSR and AODV. We introduce a systematic proce-
2
dure by which these protocols are decomposed into
their mechanistic building blocks and then evaluated
accordingly. Our results show that the contribution
of these building blocks to the overall performance
varies drastically. While flooding and caching have
a great effect on performance, route salvaging (in
DSR) barely has any effect. By using our approach
we are able to attribute performance differences be-
tween DSR and AODV to individual building blocks
under various mobility conditions. Our study in-
volves extensive simulation of carefully selected sce-
narios, and selective measurement of protocol mech-
anisms.
This is the first work, we are aware of, that takes
the building block approach to study ad hoc rout-
ing protocols, and the first to explain in detail (down
to the level of mechanisms) the difference in perfor-
mance of such protocols under various mobility pat-
terns. We hope to extend our approach to study other
ad hoc routing protocols in the future.
The rest of the paper is organized as follow. Sec-
tion II gives a brief description of our previous work.
Section III describes the building block methodology
and a procedure to identify, organize, generalize and
parameterize the building blocks. In Section IV, we
apply this methodology to the reactive MANET rout-
ing protocols. Through extensive simulations, we an-
alyze the impact of mobility on the protocol perfor-
mance in terms of building blocks in Section V. Sec-
tion VI gives a brief description of related work. Fi-
nally, our conclusions and future work are listed in
Section VII.
II. BACKGROUND
In [1], we proposed a framework to systematically
evaluate the impact of mobility on MANET routing
protocols. We defined protocol independent mobil-
ity metrics like the average degree of spatial depen-
dence (
D
spatial
) and the average relative speed (
RS)
to capture certain mobility characteristics. One of
these characteristics was the extent to which the mo-
tion of a node is influenced by nodes in its neighbor-
hood (which is captured by
D
spatial
). Another char-
acteristic was the presence of geographic restrictions
on mobility. Then, we chose mobility models that
spanned these mobility characteristics. These mobil-
ity models were:
1) Random Waypoint (RW): At every time in-
stant, a node randomly chooses a speed and
destination, and moves towards it. Each node
moves independently of other nodes.
2) Reference Point Group Mobility (RPGM):
Nodes move in either single or multiple groups.
The movement of a node in a group is strongly
influenced by the leader of the group.
3) Freeway (FW): Each node moves in its lane
on the freeway. Its movement is constrained by
nodes moving ahead of it in the same lane.
4) Manhattan (MH): Nodes move on a grid. As
in FW, each node is constrained by nodes mov-
ing ahead of it. However, unlike the FW, a node
is free to change its direction at the crosspoints
of the grid.
Different mobility patterns following the above
mobility models were generated by varying the max-
imum speed of the mobile nodes. We evaluated the
mobility metrics of these mobility patterns. Figure
2 and Figure 3 show how average degree of spatial
dependence and average relative speed vary with the
maximum speed of mobile node across various mo-
bility models. Using these mobility patterns, simu-
lations were run in the ns-2 environment for DSR,
AODV and DSDV. To explain the relationship be-
tween the mobility metrics and the protocol perfor-
mance, we defined some connectivity graph metrics.
The most useful of these graph metrics was the aver-
age link duration (
LD).
During this study, we observed that, given the
communication traffic pattern, the underlying mobil-
ity pattern does have a significant impact on the per-
formance of routing protocols. There is no clear per-
formance based ranking of the protocols. We believe
that the difference in the protocol performance comes
from the fact that the basic mechanisms (or “build-
ing blocks”) of these protocols are different. For ex-
ample, DSR and AODV are reactive while DSDV
is proactive. Even within reactive protocols, DSR
and AODV use different mechanisms and exhibit dif-
ferent performance. Thus, in [1], we conducted a
preliminary investigation on the effect of mobility
on some of the “building blocks” and their influence
on the protocol performance as a whole. In this pa-
per, we extend the previous intuitive analysis into a
thorough and quantitative analysis of the “building
blocks”.
3
III. BUILDING BLOCK METHODOLOGY
A. Motivation
Several routing protocols have been proposed for
MANETs. Based on the difference of main mech-
anisms like how to set up a route and how to react
to route failures, some protocols are categorized as
proactive [3] [6] [5] [7] [8] [9], while others are
classified as reactive [2] [4] [11] [10] [12] [13].
Within the same category, the detailed mechanisms of
the protocols vary.
Extensive research has been done to compare and
explain the differences of various routing protocols
at the whole protocol level through simulation or in-
tuitive analysis [14] [15] [16] [17]. This helped
us to rank protocols according to their performance
and distinguish the pros and cons of each protocol.
But the impact of the fine grained functionality on
the whole protocol remains unknown. We propose a
methodology to systematically compare and explain
the commonalities and differences of routing proto-
cols at a more detailed building-block level. We be-
lieve that the exact parameter settings of these build-
ing blocks and their interaction will have a significant
impact on the protocol performance under different
scenarios. In this study, we investigate the impact of
parameter settings of building blocks on the overall
performance in the face of various mobility models.
This enables us to better understand the interplay be-
tween the building blocks and mobility characteris-
tics; a topic that has been ignored in previous studies.
The thorough investigation of the behavior of
MANET protocols at the building-block level has
the following benefits:
1) We can achieve a deeper insight into the mech-
anisms of different routing protocols and the
impact of their design choices on the perfor-
mance as a whole.
2) It becomes possible to classify the several rout-
ing protocols into different categories accord-
ing to their building blocks. This can also help
in deriving generic conclusions about an entire
class (category) of protocols.
3) It may become possible to design a new proto-
col with improved performance for a given sce-
nario by combining the building blocks of dif-
ferent protocols.
B. Framework
Each protocol may be decomposed into a set of
parameterized mechanistic “building blocks”. Each
“building block” implements a specific well-defined
functionality. These building blocks are then orga-
nized in a certain way to form the protocol as a whole.
Although some protocols consist of building blocks
of the same functionality in a similar organization, it
is observed that their performance may differ signifi-
cantly. For example, AODV and DSR behave differ-
ently although they belong to the generic class of re-
active protocols. One of the reasons for this discrep-
ancy might be that the building blocks are parameter-
ized. Different values for the parameters lead to vary-
ing performance across these protocols belonging to
the same category.
Thus, the architecture of a protocol consists of
the following three elements: mechanistic building
blocks, the parameters of the mechanistic building
blocks and the interaction between the mechanistic
building blocks. For a specific mobility pattern, the
behavior of each building block is determined by
its parameter setting. At the same time, the inter-
action between building blocks affects the protocol
performance. So the performance behavior of the
whole protocol is defined by both the characteristics
of building blocks and the interaction between build-
ing blocks.
Decomposing a protocol into building blocks in a
meaningful way may not be straightforward. There
may be several ways in which this objective may be
achieved. As a first attempt, and based on our experi-
ence from previous studies, we propose a procedure
to systematically decompose routing protocols into
their constituent building blocks
. We then use the re-
active MANET routing protocols as a case study for
our procedure.
The procedure listed below is an outline to estab-
lish the building block architecture for MANET rout-
ing protocols. In section IV, we will describe how
this procedure may be applied to the reactive rout-
ing protocols as a case study. Following is a four-step
procedure by which building blocks can be identified
(by decomposition), organized, generalized and pa-
rameterized:
1) Identification of components: The whole pro-
The framework to decomposerouting protocols into building blocks
systematically and efficiently is currently under investigation. In this
paper, we mainly focus on how to build a framework that may provide
some guidelines to analyze the various protocols.
4
tocol is decomposed into several fundamental
mechanisms that perform the major functional-
ities. Hence, the first step is to identify the ma-
jor functionalities and abstract their fundamen-
tal mechanisms into building blocks.
2) Organization of components: The mechanis-
tic building blocks are linked according to their
functionalities and organized to form the whole
protocol. The links can be thought of as the
messages exchanged (or procedure calls) be-
tween these components. These links define
the interaction between the building blocks.
3) Generalization of components: Based on the
functionality of the previously chosen building
blocks, new, more generalized building blocks
may be obtained by merging mechanisms from
different protocols that have common structure
and achieve similar functionalities.
4) Parameterization of components: By changing
the setting of the parameters of these general-
ized building blocks, we can easily represent
the different mechanisms that were merged in
the previous step. Thus, each of the protocols
under study may be composed of a set of pa-
rameterized, generalized building blocks.
Applying the above procedures to reactive rout-
ing protocols, based on our experience and intuition,
we provide two simple rules of thumb to decom-
pose reactive MANET routing protocols into build-
ing blocks and organizing them:
1) Different building blocks should have
independent(non-overlapping) mechanisms.
For example, the flooding building block is
used to distribute the route requests in order to
find the appropriate route within the network.
It is used in both the route discovery and
route maintenance phases of reactive MANET
routing protocols. Thus we choose flooding
(not route discovery) as the mechanistic
building block. If we choose route discovery
as a building block and route maintenance
as another building block, then they would
overlap by using the flooding mechanism.
2) The interaction between building blocks
should be simple. The building blocks should
be identified such that most of their details
can be parameterized. Thus, a lot of detail
is hidden within each building block, while
the interaction between building blocks is
kept simple. For example, DSR uses a non-
propogating route request before flooding the
route request. We could have a separate build-
ing block for non-propogating route requests
and another for flooding. The interaction
between them occurs only when the former
is unsuccessful in finding a route. However,
both of them can be merged into a single
building block with the range of flooding as
a parameter. In the former case, the range is
1, while in the latter, the range is the network
diameter.
MANET protocols can be categorized into several
main categories: proactive, reactive, hierarchical, ge-
ographic, etc. For the sake of simplicity, as an il-
lustration, we apply the building block methodology
to reactive MANET routing protocols like DSR and
AODV. Our investigation shows that this procedure
may also be used to proactive routing protocols like
DSDV as well. A more generalized and systematic
framework to analyze the impact of design choices
on routing protocol performance based on building
blocks methodology is part of our future work.
IV. BUILDING BLOCK ANALYSIS OF REACTIVE
MANET ROUTING PROTOCOLS
Unlike the periodic message-exchange conven-
tional routing protocols, certain MANET routing pro-
tocols ( [2] [4] [11] [10] [12] [13]) adopt the source-
initiated on-demand routing approach. The distin-
guishing feature of reactive on-demand routing pro-
tocols is that the route from source to destination is
set up by the source only when required. As an il-
lustration, two well-known reactive protocols, DSR
and AODV, are analyzed based on the building block
methodology.
A. Decomposition and Organization
The operation of DSR is composed of two phases:
Route Setup phase and Route Maintenance phase.
The objecive of the former is to find a route to the
destination within the network. Two major mecha-
nisms are used to achieve this objective: A mech-
anism called global flooding is implemented to dis-
tribute the route request messages within the network
once the attempt of source to get a route reply from its
direct neighbors fails. Besides, another mechanism is
5
DSR AODV
Local Inquiry &
Global Flooding
Cache
Management
Link
Monitoring
Salvaging
Error
Notification
Expanding Ring
Search & Global
Flooding
Cache
Management
Link
Monitoring
Localized
Rediscovery
Error
Broadcast
(a) (b)
Route Setup
Route Maintenance
Flooding Caching
Range of Flooding
Caching Style
Expiration Timer
Error
Detection
Error
Handling
Error
Notification
Detection
Method
Handling
Mode
Recipient
Route Request
Add Route Cache
Route Reply
Link Breaks Notify
Route
Invalidate
Localized/Non-localized method
Notify
Generalization of
Error Handling
(c)
Generalization of
Error Handling
Generalization
of Flooding
Generalization
of Flooding
Fig. 1. Diagram of Building Block Framework for Reactive Protocols
implemented to manage the cached routing informa-
tion at the nodes, including how to add, invalidate and
utilize the cached route entries.
Unlike the conventional wired network, the links
within the wireless network are highly unstable due
to the mobility and wireless propagation losses. The
unstable links are dealt with in the Route Mainte-
nance phase. DSR monitors the link status at the
MAC layer. If a broken link is detected, the salvaging
mechanism is used to find an alternative route. At
the same time, in DSR, a route error message is sent
in the direction of the source to eliminate the invalid
cache entries.
The operation of AODV is similar to DSR, the
routing function is achieved in Route Setup phase and
Route Maintenance phase: Expanding ring search
and global flooding schemes are in charge of dis-
tributing the route request message in the network
while a caching mechanism is used to maintain routes
and to reply the route requests based on the cached
routing information. Hello messages are used by
AODV to monitor the link status. If a broken link
is detected, a localized route discovery mechanism is
re-initiated by the upstream node to repair the broken
route in some scenarios. Nodes within the network
are notified about the error so that the stale cache en-
tries can be removed.
The part (a) and part (b) of Figure 1 show the ar-
chitecture for DSR and AODV, respectively, based on
the BRICS approach. This kind of architecture is one
of many possible candidates
.
B. Generalization and Parameterization
From the previous section, we observe that both
DSR and AODV can be decomposed into a similar set
of basic mechanistic building blocks. By comparing
and analyzing the commonalities and differences of
the functionalities of these components, we can con-
struct generalized building blocks that encompasses
the building blocks of each protocol. The part (c) of
Figure 1 shows a generalized building block archi-
tecture for both DSR and AODV. This architecture is
obtained by combining the common mechanisms of
the two protocols into building blocks. As a exam-
ple shown in Figure 1, salvaging from DSR and lo-
calized error recovery from AODV are compared and
merged as the error handling building block whose
major functionality is to find alternative route to re-
place the invalid route. The parameter of this gen-
Based on our study, we find that the interaction between the build-
ing blocks can be very complex dependingon different conditions. The
nature of these interactions is currently under investigation
6
eralized building block is the mode of error han-
dling used. Similarly, local inquiry and global flood-
ing from DSR and expanding ring search and global
flooding from AODV can be merged into the flooding
building block with the TTL as a parameter.
Once the building blocks are generalized, the key
parameters of building blocks can be identified by
studying the difference in the functionality and be-
havior of the same building block in various pro-
tocols. We now discuss the design choices (pa-
rameter settings) of each of the identified building
blocks.
1) Route Setup Phase: In this phase, the flood-
ing building block takes responsibility to dis-
tribute the route request messages within the
network. The caching building block helps to
efficiently and promptly provide the route to
the destination without referring to the destina-
tion every time.
Flooding: Flooding is mainly used for route
discovery if the route to the destination does
not exist in the sender’s cache. Here, the key
parameter is the range of flooding, generally
described by TTL field in the IP header. If the
TTL is set to the network diameter, a global
flooding is done. One optimization is to im-
plement the localized controlled flooding
be-
fore global flooding. This may be useful if the
probability of finding an appropriate cache in
the neighborhood is high.
Caching: Several parameters affect the behav-
ior of the caching building block. One pa-
rameter is whether aggressive caching is al-
lowed, i.e. whether multiple cache entries are
allowed for the same destination and whether
a node can cache the route information it over-
hears? Generally speaking, aggressive caching
scheme increases the possibility of finding an
appropriate route without re-initiating a route
discovery. Another parameter is the expira-
tion timer for route (cache) entry. The ex-
isting cache entry may become stale due to
node movement. Hence, a reasonable set-
ting of the expiration timer will keep the fresh
cached routes valid as well as eliminate the
stale cached routes.
2) Route Maintenance Phase: This phase is
In localized controlled flooding, the route request may be sent in a
smaller neighborhood i.e. within 1 hop, 2 hops and so on.
made up of the following building blocks:
Error Detection: It is used to monitor the sta-
tus of the link of a node with its immediate
neighbors. Several methods to monitor the link
status between neighbors can be used: MAC
level acknowledgement, network-layer explicit
Hello message or network-layer passive over-
hearing scheme. Here, the parameter is the
mode of error detection used.
Error Handling: It finds alternative routes to
replace an invalid route after a broken link is
detected. One of the parameters to this block is
whether localized recovery should be used.Ina
non-localized recovery, the node detecting the
link breakage will notify the source to handle
the error. The source will re-initiate the route
discovery procedure if the route is still needed.
In a localized recovery, the node detecting the
broken link will attempt to find an alternative
route in its own cache or do a localized flooding
before asking the source to re-initiate the route
discovery.
Error Notification: It is used to notify the
nodes in the network about invalid routes. The
key parameter to this building block is the re-
cipient of the error message. Either only the
source is notified or the entire network is noti-
fied.
C. Comparison of parameters of DSR and AODV
Having identified the generalized building blocks,
we now study the specific parameter settings for
these building blocks for DSR and AODV. We pose
some questions about the utility of the various design
choices made by these protocols. In section V, we at-
tempt to answer these questions.
1) Flooding: For the range of flooding, both DSR
and AODV use the two-step controlled flood-
ing. DSR conducts a non-propagating direct-
neighborhood inquiry(TTL=1) first before the
global flooding(TTL=D, D is network diam-
eter). Similarly, AODV
uses the expand-
ing ring search(TTL=1,3,5,7) before the global
flooding is initiated. Here, we want to answer
the following question: How useful are non-
propogating route requests?
Current AODV implementation in (ns-2)(version ns-2.1b8a) adopts
the expanding ring search, although the original AODV paper [4] uses
only the global flooding.
7
2) Caching: DSR uses aggressive caching, while
AODV does not. Another parameter is the ex-
piration timer for the cache entry. Classifica-
tion of the different expiration timers and their
effect on the performance can be a study in it-
self [20]. We will investigate this parameter in
detail in the future. For caching, We are inter-
ested in the following questions: How useful is
caching? and Is aggressive caching better than
non-aggressive caching?
3) Error Detection: For mode of error detection,
both DSR and AODV can use either of the
three choices mentioned in section IV-A. Since
all these schemes are very similar, we do not in-
vestigate this building block in our analysis
.
4) Error Handling: For localized recovery,in
DSR, on detecting a broken link, the upstream
node will first search its cache to replace the
invalid route, although the found alternative
route may also be invalid in some scenarios.
While in AODV, the upstream node detecting
the broken link will initiate a localized flood-
ing to find the route to the destination
. For this
building block, we are interested in the follow-
ing question: Which is a better scheme for lo-
calized error handling: cache lookup or local-
ized flooding?
5) Error Notification: For recepient of error noti-
fication, both DSR and AODV
notify the error
to the source. Since both DSR and AODV use
the same parameter setting, we do not investi-
gate this block during our simulations.
V. PERFORMANCE EVALUATION AND
DISCUSSION
In this section, we are interested in answering the
design questions for the following building blocks as
mentioned in section IV-C:
1) Flooding
2) Caching
3) Error Handling
However, the interaction of this building block with the others may
impact the performance. We will study it in more detail when we in-
vestigate the interactions between the building blocks
Current AODV implementation in (ns-2)(version ns-2.1b8a) uses
this scheme, while the original AODV paper [4] only allows the non-
localized error handling.
Current AODV implementation in (ns-2)(version ns-2.1b8a) noti-
fies the error to the source, while the original AODV paper [4] allows
the upstream node to notify all the nodes having this route entry.
We find that although the mechanisms used in DSR
and AODV are quite similar, the exact parameter set-
tings of these mechanisms can have a significant im-
pact on the protocol performance in the face of mobil-
ity. We identified parts of the network simulator (ns-
2) code which implement these building blocks and
profiled them during our simulations [21].
We carried out simulations in the ns-2 environment
with the CMU Wireless Ad Hoc networking exten-
sion. The transmission range of the nodes was 250m.
Our mobility scenario generator produced the differ-
ent mobility patterns following the RPGM, Freeway
(FW) and Manhattan (MH) models according to the
format required by ns-2. In all these patterns, 40 mo-
bile nodes moved in an area of 1000m x 1000m for
a period of 900 seconds. Random Waypoint (RW)
mobility pattern was generated using the setdest tool
which is a part of the ns-2 distribution. For RPGM,
we used 2 different mobility scenarios: single group
of 40 nodes and 4 groups of 10 nodes each moving in-
dependently of each other and in an overlapping fash-
ion. Both Speed Deviation Ratio (SDR) and Angle
Deviation Ratio (ADR) were set to 0.1
. For the Free-
way and Manhattan models, the nodes were placed on
the freeway lanes or local streets randomly in both di-
rections initially. Their movement was controlled as
per the specifications of the models. The maximum
speed V
max
was set to 1, 5, 10, 20, 30, 40, 50 and 60
m/sec to generate different movement patterns for the
same mobility model.
The traffic pattern was generated by the cbrgen tool
that is part of the ns-2 distribution. The traffic con-
sisted of 20 Constant Bit Rate (CBR) sources and 30
connections. The source destination pairs were cho-
sen at random. The data rate used was 4 packets/sec
and the packet size was 64 bytes.
A. Analysis of DSR and AODV
For the remaining part of the paper, we will refer
to the cache as some sort of a collective cache i.e. a
cache that consists of the cache entries of all the nodes
in the simulation.
1) Flooding: In our experiments, we find that
the numbers of route requests for both AODV and
DSR increase as the mobility increases, Figures 4 and
5 show that the number of route requests increase
with V
max
in a trend similar to the mobility metrics
SDR and ADR are defined in [1]. They control the extent to which
the group members can deviate from the leader in speed and direction.
8
shown in Figure 2. This observation indicates that
the behavior of building blocks, which in turn affects
the performance of routing protocols, is strongly cor-
related with the mobility metric of average relative
speed
. This observation indicates that mobility does
affect the performance of routing protocols as con-
cluded in [1].
Figures 6 and 7 show the ratio of non propagating
route requests to the total number of route requests is-
sued by the DSR and AODV respectively. This met-
ric measures the likelihood of finding a route to the
destination from the source’s neighbors.
It is observed that this ratio increases from RW to
MH to FW mobility models. This is because the ge-
ographic constraints on movement are greater in FW
than MH, which are in turn greater than in RW. Thus,
the likelihood of finding a route to the destination
from the source’s neighbors increases from RW to
MH to FW. RPGM (Single and Multiple groups) has
a high average degree of spatial dependence. Hence,
it is expected that this ratio will be high. However it
turns out that this is not the case. As figure 7 shows,
the ratio is the lowest for RPGM. This is because,
in RPGM, the propagating route requests make up a
large percentage of the total route requests. These
propagating route requests are issued when the source
sets up a connection to the destination for the first
time. Since the route remain stable for long periods of
time, very few route requests retries are done by the
source. In DSR, we find that RPGM (with 4 groups)
has the highest ratio, whereas it should have been at
the same level as RPGM (with single group). This
discrepancy is because of the smaller number of route
requests issued by DSR vis-a-vis AODV as shown
by figures 4 and 5. In order to avoid the unnecessary
confusion, we remove the corresponding curve from
Figure 6.
On the other hand, the ratio for DSR is almost
twice as large as that for AODV across all mobil-
ity models. A possible reason for this might be the
fact that DSR uses aggressive caching as compared to
AODV. When such a caching scheme is coupled with
the mechanism of non propagating route requests, it
translates to low routing overhead and high through-
put as was shown in [1] and several other compara-
tive studies.
In experiments, we find that several other metrics of building
blocks, such as Total Number of Route Replies, Number of Route
Replies from the Cache, Number of Route Replies from the Destination,
Number of Broken Links, Number of Route Error Messages etc. are
strongly correlated with the mobility metric of average relative speed.
0 10203040 50 60
Maximum Speed (m/sec)
0
10
20
30
40
50
Average Relative Speed (m/sec)
Random Waypoint
RPGM (Single Group)
RPGM (4 Groups)
Freeway
Manhattan
Fig. 2. Average Relative Speed
0 10203040 50 60
Maximum Speed (m/sec)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.55
Average Degree of Spatial Dependence
Random Waypoint
RPGM (Single Group)
RPGM (4 groups)
Freeway
Manhattan
Fig. 3. Average Degree of Spatial Dependence
Thus, it seems that caching has a significant impact
on the performance of DSR and AODV. Hence we
study it next.
2) Caching: To measure the effectiveness of
caching, we evaluate the ratio of the number of route
replies coming from the cache to the total number of
route replies.
Figures 8 and 9 show that this ratio is high for
RW, MH and FW models, which implies that most
of the route replies for these mobility models come
from the cache. On the other hand, since RPGM (Sin-
gle and Multiple Groups) has a high degree of spa-
tial dependence, it would be expected that most of the
route replies can be found in the cache of the nearby
nodes. However, it turns out that the ratio is lower for
RPGM. This is because, as mentioned in section V-
A.1, most of the route replies come from the destina-
tion when the first route request is sent by the source
to the destination. After this, the route remains sta-
ble during the entire simulation and thus very few
route requests retries are done. RPGM with multiple
groups has a lower degree of spatial dependence as
9
0 10203040 50 60
Maximum Speed (m/sec)
0
2000
4000
6000
8000
Total Number of Route Requests
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
RPGM (4 Groups)
Fig. 4. Number of Route Requests issued by DSR
0 10203040 50 60
Maximum Speed (m/sec)
0
5000
10000
15000
20000
Total Number of Route Requests
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
RPGM (4 Groups)
Fig. 5. Total Number of Route Requests issued by AODV
0 10203040 50 60
Maximum Speed (m/sec)
0
0.2
0.4
0.6
0.8
# of Non-Propogating Route Reqs. / Total # of Route Reqs.
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
Fig. 6. Ratio of Non Propogating Route Requests to Total Number of
Route Requests for DSR
0 10203040 50 60
0
0.1
0.2
0.3
0.4
Random Waypoint
Manhattan
Freeway
RPGM (1 Group)
RPGM (4 Groups)
Fig. 7. Ratio of Non Propogating Route Requests to Total Number of
Route Requests for AODV
0 10203040 50 60
Maximum Speed (m/sec)
0
0.2
0.4
0.6
0.8
1
# Route Replies from Cache / Total # of Route Replies
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
RPGM (4 Groups)
Fig. 8. Ratio of the Number of Route Replies from the Cache to the
Total Number of Route Replies for DSR
compared to the single group case. Thus, the stabil-
ity of the routes in the case of multiple groups is lesser
than that in the single group case. Hence, the ratio is
higher in the former case.
The difference in the ratio for DSR and AODV is
greater than 0.2 for all mobility models. DSR uses
aggressive caching as compared to AODV. Thus, the
likelihood of a route reply coming from a cache is
higher in DSR than in AODV. Thus, fewer route re-
quests will be needed and thus the routing overhead
of DSR is lower than AODV as was concluded in sev-
eral comparative studies. Thus, aggressive caching
seems to be a good design choice.
To completely evaluate the caching strategy, we
also need to examine the validity of the cache entries.
We evaluate the ratio of invalid cache entries to the
total number of cache entries for DSR.
As shown by Figure 10, the ratio increases from
RPGM to RW to FW to MH mobility models. Also,
10
0 10203040 50 60
Maximum Speed (m/sec)
0.2
0.4
0.6
0.8
1
# of Route Replies from the Cache / Total # of Route Replies
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
RPGM (4 Groups)
Fig. 9. Ratio of the Number of Route Replies from the Cache to the
Total Number of Route Replies for AODV
0 10203040 50 60
Maximum Speed (m/sec)
0
0.2
0.4
0.6
0.8
1
# of Invalid Cache Replies / Total # of Cache Replies
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
RPGM (Multiple Groups)
Fig. 10. Ratio of the Number of Invalid Route Replies from the Cache
to the Total Number of Route Replies from the Cache
in section V-A.2, it was seen that the ratio of route
replies from the cache to the total number of route
replies was higher for MH than FW which in turn was
higher than RW. Thus caching may have adverse ef-
fects in mobility models with a high relative speed.
Packets may be sent on invalid routes which might
lead to packets being dropped and route request re-
tries. This leads to a lower throughput and higher
overhead for DSR for the RW, FW and MH models
as was shown in [1]

.
On the other hand, in mobility models with very
high relative speed like MH and FW, AODV seems to
achieve as good a throughput as DSR (and sometimes
We evaluated this ratio only for DSR, but extended the conclusion
to AODV too. We believe that a similar effect causes performance
degradation in AODV . However, in AODV entire routes are not cached.
So, it is harder to find if a route is invalid. We suspect that in AODV ,
packets might be getting dropped due to TTL hitting zero. Evaluating
such losses is currently under investigation
0 10203040 50 60
Maximum Speed (m/sec)
0
0.1
0.2
0.3
0.4
0.5
# of Localized Route Requests / Total # of Route Errors
Random Waypoint
Manhattan
Freeway
RPGM (Single Group)
RPGM (4 Groups)
Fig. 11. Ratio of the Number of Localized Route Recovery Requests
to the Total Number of Route Errors for AODV
better) as was shown by [1]. AODV does not use ag-
gressive caching, thus the ratio of the number of route
replies coming from the cache to the total number of
route replies is lesser for AODV than DSR. Thus, the
likelihood of getting invalid routes from the cache is
lesser for AODV than for DSR. Moreover, at high rel-
ative speeds, the number of routes broken is greater.
Thus, a protocol which has a better error handling
mechanism at higher relative speeds might perform
better in such situations. This line of reasoning leads
us to evaluate the next building block of interest - Er-
ror Handling.
3) Error Handling: To study the effectiveness
of error handling, we focus on localized error
handling

. We evaluate the ratio of the number of
localized error handling to the total number of route
errors for both DSR and AODV. For, DSR, we notice
that salvaging accounts for less than 2% of the total
number of route errors. Moreover, if we take invalid
cache entries into account, the effect of salvaging on
the protocol performance is further lowered. On the
other hand, in AODV, a route request is initiated by
the upstream node which detects the broken link if it
is closer to the destination. As figure 11 shows, the
ratio is between 0.4 - 0.5 for FW and MH models.
Moreover the routes obtained by this mechanism are
more up to date than those from the cache. This is
probably another factor which explains the better
performance of AODV as compared to DSR in the
FW and MH models.
Non-localized error handling is done by the source for both DSR
and AODV . Hence we do not study the non localized error handling as
they are the same for both the protocols
11
B. Discussion
The above study of the building blocks has given
us greater insight into the design of the reactive rout-
ing protocols for MANETs. Decomposing a proto-
col into building blocks and evaluating these building
blocks have shown us the scenarios in which the cho-
sen parameters can give a better performance. From
the above study, we learnt the following principles of
protocol design:
1) Caching helps reduce the protocol overhead.
However, whether aggressive caching should
be used depends on the scenarios in which the
protocol will be deployed. For low mobility
scenarios, aggressive caching might be useful,
while for higher mobility scenarios, the stale
cache entries might affect the protocol through-
put.
2) Non Propogating route requests, when com-
bined with caching also reduce the protocol
overhead. If caching is widely done in the
network, it may be more advantageous to do
non propagating route requests (or expanding
ring search) than globally flooding the route
request. In DSR, due to aggressive caching,
it may be more useful to do expanding ring
search (from the source) on a route error than
doing a global flooding (from the source).
Again this might work well only for low mo-
bility scenarios.
3) The nature of localized error handling also has
a significant impact on protocol performance.
Re-initiating a route request from an intermedi-
ate node can be more advantageous than doing
a local cache lookup in high mobility scenarios,
while a cache lookup might be more advanta-
geous for low mobility scenarios.
Thus, no particular parameter setting of these
building blocks is the most optimal for all scenarios.
This further strengthens our conclusion in [1] that
there is no clear winner among the protocols across
all mobility scenarios.
VI. RELATED WORK
To the best of our knowledge, no work has been
done to compare the performance of MANET rout-
ing protocols and explain the difference based on the
building block behavior. However, we draw inspira-
tion from some related work in this area.
Extensive work has been done to compare the per-
formance of several routing protocols by [14], [15],
[16] and [17]. Their focus has been comparison
of routing protocol mechanisms and the explanation
of the difference between various protocols at the
“whole protocol” level.
In [18], several design choices of DSR were care-
fully analyzed through simulations and some impor-
tant observations were provided to help better un-
derstand the unique on-demand nature of DSR. The
performance and mechanism of two reactive routing
protocols, DSR and AODV, were compared and dis-
cussed in [16]. However, we attempt to provide a
systematic framework to decompose a protocol into
mechanistic building blocks to analyze the impact of
the numerous design choices available.
In [1], we proposed the building block framework
as a potential methodology to study the protocol be-
havior. We used DSR as an example to show how this
framework can be applied to explain the protocol be-
havior under different mobility scenarios. In this pa-
per, we develop the building block framework. We
demonstrate the utility of the framework by bringing
out the effect of various design choices on the perfor-
mance of procotols, particularly DSR and AODV.
VII. CONCLUSION AND FUTURE WORK
From previous studies, we observed that perfor-
mance of MANET routing protocols vary signifi-
cantly with mobility. Moreover, it was observed
that mobility impacts different protocols differently
even if they follow similar mechanisms as in the case
of DSR and AODV. In an attempt to answer such
performance discrepancies, we carried our research
beyond the whole protocol level. This led to the
building block based approach for analyzing rout-
ing protocols. Using this approach, we analyze the
reactive MANET routing protocols as a case study.
We demonstrate the possible benefits of such an ap-
proach by relating the performance of the building
blocks to the protocol performance. Through ex-
periments, several lessons about protocol design are
learnt. These lessons may help in designing protocols
with better performance across several mobility mod-
els (or other conditions).
As part of the future work, we would like to ex-
tend this approach to pro-active protocols like DSDV,
geographic routing protocols, among others. While
proposing the building block architecture, we used
12
our intuition about the mechanisms of DSR and
AODV. To accomplish the same in a generalized
manner, using an algorithm, would be extremely
challenging. Moreover, we would like to focus on an-
alyzing the interactions between the building blocks
and their effects on the protocol performance.
REFERENCES
[1] F. Bai, N. Sadagopan and A. Helmy. IMPORTANT: A frame-
work to systematically analysis the Impact of Mobility on Per-
formance of RouTing protocols for Adhoc NeTwork. accepted
to IEEE INFOCOM 2003.
[2] D.B.Johnson, D.A.Maltz and J. Broch. DSR:The Dynamic
Source Routing Protocol for Multi-Hop Wireless Ad Hoc Net-
work. In Ad Hoc Networking, edited by C.E.Perkins, Chapter
5, pp.139-172, Addison-Wesley, 2001.
[3] C.E.Perkins and P. Bhagwat. Highly dynamic destination se-
quenced distance vector routing (DSDV) for mobile comput-
ers, in ACM SIGCOMM, pp 234,244, 1994.
[4] C. Perkins. Ad hoc On Demand Distance Vector Rout-
ing(AODV), internet draft, draft-ietf-aodv-00.txt
[5] S. Murthy and J.J. Garcia-Luna-Aceves. An Efficient Routing
Protocol for Wireless Networks, ACM Mobile Networks and
App. J., Special Issue on Routing in Mobile Communication
Networks, Oct. 1996, pp. 183-97.
[6] S. Murthy, J.J. Garcia-Luna-Aveces. Distributed Bellman-
Ford routing protocol (DBF), A Routing Protocol for Packet
Radio Networks, Proc. ACM International Conference on Mo-
bile Computing and Networking, pp. 86-95, November, 1995.
[7] A. O’Nell, H. Li. Hiearchical State Routing Protocol Internet
Draft, draft-oneill-li-hsr-00.txt.
[8] P. Jacquet, P. Muhlethaler, A. Qayyum, A. Laquiti, L. Vien-
not, T. Clausen. Optimized Link State Routing Protocol Inter-
net Draft, draft-ietf-manet-olsr-04.txt, work in progress, June
2001.
[9] T-W. Chen and M. Gerla. Global State Routing: A New
Routing Scheme for Ad-hoc Wireless Networks. Proc. IEEE
ICC’98, 5 pages.
[10] V . Park, S. Corson. Temporally-ordered Routing Algorithm
(TORA) VERSION 1 Internet Draft, draft-ietf-manet-tora-
spec- 03.txt, work in progress, June 2001.
[11] C-K. TOH. Associativity-based long-lived Routing(ABR)
Protocol, Internet Draft.
[12] R. Dube, C. D. Rais, K. Wang and S. K. Tripathi. Signal Sta-
bility based adaptive routing (SSR alt SSA) for ad hoc mobile
networks, IEEE Personal Communication, Feb. 1997.
[13] M.S. Corson and A. Ephremides. Lightweight Mobile Rout-
ing protocol (LMR) ,A distributed routing algorithm for mo-
bile wireless networks, Wireless Networks 1 (1995).
[14] J. Broch, D. A. Maltz, D. B. Johnson, Y-C. Hu, and J. Jetcheva.
A Performance Comparison of Multi-Hop Wireless Ad Hoc
Network Routing Protocols. In Proceedings of the Fourth An-
nual ACM/IEEE International Conference on Mobile Comput-
ing and Networking, ACM, Dallas, TX, October 1998.
[15] S-J. Lee, M. Gerla, and C-K. Toh.A Simulation Study of
Table-Driven and On-Demand Routing Protocols for Mobile
Ad Hoc Networks. IEEE Network,vol. 13, no. 4, Jul/Aug
1999, pp. 48-54.
[16] S. R. Das, C. E. Perkins and E. M. Royer. Performance Com-
parison of Two On-demand Routing Protocols for Ad Hoc Net-
works. Proceedingsof INFOCOM 2000 Conference, Tel-Aviv,
Israel, March 2000.
[17] S. R. Das, R. Castaneda, J. Yan and R. Sengupta. Comparative
Performance Evaluation of Routing Protocols for Mobile, Ad
hoc Networks. In Proceedings of 7th Int. Conf. on Computer
Communications and Networks (IC3N), Lafayette, LA, Octo-
ber, 1998, pages 153-161.
[18] D. A. Maltz, J. Broch, J. Jetcheva, and D. B. Johnson. The Ef-
fects of On-Demand Behavior in Routing Protocols for Multi-
Hop Wireless Ad Hoc Networks. IEEE Journal on Selected Ar-
eas in Communications special issue on mobile and wireless
networks. August 1999.
[19] E. M. Royer and C-K. Toh. A Review of Current Routing Pro-
tocols for Ad-Hoc Mobile Wireless Networks. IEEE Personal
Communications Magazine, April 1999, pp. 46-55.
[20] Y-C. Hu and D. B. Johnson. Caching Strategies in On-Demand
Routing Protocols for Wireless Ad Hoc Networks. Proceed-
ings of the Sixth Annual ACM/IEEE International Conference
on Mobile Computing and Networking (MobiCom 2000),
ACM, Boston, MA, August 2000.
[21] L. Breslau, D. Estrin, K. Fall, S. Floyd, J. Heidemann, A.
Helmy, P. Huang, S. McCanne, K. Varadhan, Y . Xu and H.
Yu, Advances in Network Simulation, IEEE Computer, vol.
33, No. 5, p. 59-67, May 2000.

Linked assets

Computer Science Technical Report Archive

Conceptually similar

PDF

USC Computer Science Technical Reports, no. 765 (2002)

PDF

USC Computer Science Technical Reports, no. 789 (2003)

PDF

USC Computer Science Technical Reports, no. 816 (2004)

PDF

USC Computer Science Technical Reports, no. 778 (2002)

PDF

USC Computer Science Technical Reports, no. 674 (1998)

PDF

USC Computer Science Technical Reports, no. 749 (2001)

PDF

USC Computer Science Technical Reports, no. 698 (1999)

PDF

USC Computer Science Technical Reports, no. 780 (2002)

PDF

USC Computer Science Technical Reports, no. 860 (2005)

PDF

USC Computer Science Technical Reports, no. 809 (2003)

PDF

USC Computer Science Technical Reports, no. 790 (2003)

PDF

USC Computer Science Technical Reports, no. 730 (2000)

PDF

USC Computer Science Technical Reports, no. 803 (2003)

PDF

USC Computer Science Technical Reports, no. 690 (1998)

PDF

USC Computer Science Technical Reports, no. 806 (2003)

PDF

USC Computer Science Technical Reports, no. 884 (2006)

PDF

USC Computer Science Technical Reports, no. 727 (2000)

PDF

USC Computer Science Technical Reports, no. 726 (2000)

PDF

USC Computer Science Technical Reports, no. 797 (2003)

PDF

USC Computer Science Technical Reports, no. 856 (2005)

Description

Fan Bai, Narayanan Sadagopan, Ahmed Helmy. "A building-block approach for analyzing routing protocols in ad hoc networks - a case study of reactive routing protocols." Computer Science Technical Reports (Los Angeles, California, USA: University of Southern California. Department of Computer Science) no. 775 (2002).

Asset Metadata

Creator Bai, Fan (author), Helmy, Ahmed (author), Sadagopan, Narayanan (author)

Core Title USC Computer Science Technical Reports, no. 775 (2002)

Alternative Title A building-block approach for analyzing routing protocols in ad hoc networks - a case study of reactive routing protocols (title)

Publisher Department of Computer Science,USC Viterbi School of Engineering, University of Southern California, 3650 McClintock Avenue, Los Angeles, California, 90089, USA (publisher)

Tag OAI-PMH Harvest

Format 12 pages (extent), technical reports (aat)

Language English

Unique identifier UC16269743

Identifier 02-775 A Building-block approach for analyzing Routing protocols in ad hoc networks - a Case Study of reactive routing protocols (filename)

Legacy Identifier usc-cstr-02-775

Format 12 pages (extent),technical reports (aat)

Rights Department of Computer Science (University of Southern California) and the author(s).

Internet Media Type application/pdf

Source 20180426-rozan-cstechreports-shoaf (batch), Computer Science Technical Report Archive (collection), University of Southern California. Department of Computer Science. Technical Reports (series)

Access Conditions The author(s) retain rights to their work according to U.S. copyright law. Electronic access is being provided by the USC Libraries, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name USC Viterbi School of Engineering Department of Computer Science

Repository Location Department of Computer Science. USC Viterbi School of Engineering. Los Angeles\, CA\, 90089

Repository Email csdept@usc.edu

Inherited Values

Title Computer Science Technical Report Archive

Description Archive of computer science technical reports published by the USC Department of Computer Science from 1991 - 2017.

Coverage Temporal 1991/2017

Repository Email csdept@usc.edu

Repository Name USC Viterbi School of Engineering Department of Computer Science

Repository Location Department of Computer Science. USC Viterbi School of Engineering. Los Angeles\, CA\, 90089

Publisher Department of Computer Science,USC Viterbi School of Engineering, University of Southern California, 3650 McClintock Avenue, Los Angeles, California, 90089, USA (publisher)