Baltzer Journals
An Ecient Routing Proto col for Wireless Networks
Shree Murthy and J.J. Garcia-Luna-Aceves
Computer Engineering, University of California,
Santa Cruz, CA 95064
We present the wireless routing proto col (WRP). In WRP, routing nodes commu-
nicate the distance and second-to-last hop for each destination. WRP reduces the
number of cases in which a temporary routing lo op can o ccur, which accounts for
its fast convergence properties. A detailed proof of correctness is presented and its
performance is compared by simulation with the p erformance of the distributed
Bellman-Ford algorithm (DBF), DUAL (a lo op-free distance-vector algorithm)
and an ideal link-state algorithm (ILS), which represent the state of the art of
internet routing. The simulation results indicate that WRP is the most ecient
of the alternatives analyzed.
Keywords
: routing, packet radio, wireless, distance-vector, link-state, shortest-
path
1{Introduction
The routing protocols used in multihop packet-radio networks implemented in
the past [2,3,11]were based on shortest-path routing algorithms that have b een
typically based on the distributed Bellman-Ford algorithm (DBF) [4]. According
to DBF, a routing node knows the length of the shortest path from each neighbor
to every network destination and this information is used to compute the shortest
path and successor in the path to each destination. An update message contains a
vector of one or more entries, each of which species as a minimum, the distance to
a given destination. A ma jor performance problem with DBF is that it takes a very
long time to up date the routing tables of network nodes after network partitions,
node failures, or increase in network congestion. This p erformance problem of
DBF stems from the fact that it has no inherent mechanism to determine when a
network no de should stop incrementing its distance to a given destination. This
problem is usually called the
counting-to-innity
problem.
The counting-to-innity problem is overcome in one of three ways in existing
This work was supp orted in part by the Advanced Research Pro jects Agency (ARPA) under
contract F19628-93-C-0175 and by the Oce of Naval Research under Contract No. N-00014-
92-J-1807.
S. Murthy and J.J. Garcia-Luna
2
internet routing protocols. OSPF [12] relies on broadcasting complete topology
information among routers, and organizes an internet hierarchically to cope with
the overhead incurred with topology broadcast. BGP [16] exchanges distance
vectors that sp ecify complete paths to destinations. EIGRP [1] uses a lo op-free
routing algorithm called DUAL [8], which is based on internodal coordination
that can span multiple hops; DUAL also eliminates temporary routing lo ops.
However, there are signicant dierences b etween wireless networks and wired
internets in whichinternet routing protocols are used. A wired internet has rela-
tively high bandwidth and top ology that changes infrequently; in contrast, wire-
less networks have mobile nodes and have limited bandwidth for network control.
Accordingly, ooding, multihop internodal synchronization and the specication
of complete path information would incur too muchoverhead in a multihop ra-
dio network with a dynamic topology. On the other hand, the routing protocols
based on DBF or mo dications of DBF would take a long time to converge and the
frequent topology changes in a wireless network with mobile no des make the loop-
ing problem of DBF unacceptable. Therefore, there is a need for a new routing
protocol whichisdevoid of all these drawbacks.
In the recent past, a number of eorts have been made to address the limitation
of DBF and topology broadcast in mobile wireless networks. One such eort is
the DSDV protocol [14]. In this proto col, each mobile host, which is a sp ecialized
router that periodically advertises its view of the interconnection topology with
other mobile hosts within the network to maintain up to date information ab out
the status of the network. Unfortunately, in DSDV a node has to wait until it
receives the next up date message originated by the destination in order to update
its distance-table entry for that destination. This implicit destination-centered
synchronization suers from the same latency problems of DUAL and similar
algorithms based on explicit synchronization. Also, DSDV uses b oth perio dic and
triggered up dates for updating routing information, which could cause excessive
communication overhead.
A distributed routing algorithm for mobile wireless networks based on diusing
computations has been proposed by Corson and Ephremides [6]. This protocol
relies on the exchange of short control packets forming a
query-reply
process. It
also has the ability to maintain multiple paths to a given destination. This is
a destination-oriented proto col in which separate versions of the algorithm run
independently for each destination. Routing is source-initiated, which means that
routes are maintained by those sources which actually desire routes. Even though
this algorithm provides multiple paths to the destination, because of the query-
based synchronization approachtoachieve lo op-free paths, the communication
complexity could b e high.
Recently,anumber of distributed shortest-path algorithms have b een pro-
posed [5, 7, 9, 10,15] that utilize information regarding the length and second-
S. Murthy and J.J. Garcia-Luna
3
to-last hop (predecessor) of the shortest path to each destination to eliminate
the counting-to-innity problem of DBF. We call this type of algorithms as
path-
nding algorithms
. According to these algorithms, each node maintains the shortest-
path spanning tree rep orted by its neighbors. A node uses this information along
with the cost of adjacent links to generate its own shortest-path spanning tree.
An update message exchanged among neighbors consists of a vector of entries
that report updates to the sender's spanning tree; each up date entry contains a
destination identier, the distance to the destination, and the second-to-last hop
of the shortest path to the destination.
Path-nding algorithms are an attractive approach for wireless networks, be-
cause they eliminate counting-to-innity problem. However, these algorithms can
still incur temp orary lo ops in the paths specied by the predecessor before they
converge; without prop er precautions, this can lead to slow convergence, or incur
substantial processing if a node is required to update its entire routing table for
each input event. To address these problems, wehave prop osed a path-nding
algorithm, PFA, which substantially reduces temporary lo oping situations [13],
and which limits routing table up dates to include only that entries aected bya
network change.
The rest of this pap er describes a wireless routing proto col (WRP) for a packet
radio network based on PFA, illustrating the key asp ects of the proto col's opera-
tion. The following sections show that the protocol is correct (i.e., that it produces
correct routing tables within a nite time after topology changes) and compares
its performance with that of DBF, DUAL and an ideal link state algorithm (ILS)
which uses Dijkstra's shortest path algorithm.
ILS consists of ideal oo ding of link-state up dates in order to replicate the
topology of the network at each router; ideal o oding means that innite sequence
numbers can b e used to validate link-state up dates, and that all such up dates are
successfully delivered at every router.
2 { Wireless Routing Proto col
2.1 { Overview
To describ e WRP,we model a network as an undirected graph represented as
G
(
V; E
), where
V
is the set of no des and
E
is the set of links (or edges) connecting
the no des. Each node represents a router and is a computing unit involving a
processor, lo cal memory and input and output queues with unlimited capacity.
In a wireless network, a node has radio connectivity with multiple no des and a
single physical radio link connects a no de with many other no des. However, for
the purposes of routing-table updating, a node
A
can consider another node
B
to be adjacent(we call such a node a \neighbor") if there is radio connectivity
S. Murthy and J.J. Garcia-Luna
4
between
A
and
B
and
A
receives up date messages from
B
. Accordingly,we map
aphysical broadcast link connecting multiple nodes into multiple point-to-point
functional links dened for these no de paths that consider to be neighbors of each
other.
Then, a functional bidirectional link connecting the nodes is assigned a p osi-
tiveweight in each direction. All messages received (transmitted) byanode are
put in an input (output) queue and are pro cessed in FIFO order. The communi-
cation links in the network are such that all up date messages transmitted over an
operational link are received in the order in which they were transmitted within
a nite time.
A link is assumed to exist between two nodes only if there is radio connectivity
between the two nodes and they can exchange update messages reliably with a
certain probability of success. When a link fails, the corresp onding distance entries
in a no de's distance and routing tables are marked as innity. A no de failure is
modeled as all links incident on that node failing at the same time.
WRP is designed to run on top of the medium-access control protocol of a
wireless network. Update messages may be lost or corrupted due to changes
in radio connectivity or jamming. Reliable transmission of up date messages is
implemented by means of retransmissions. After receiving an update message free
of errors, a no de is required to send a p ositiveacknowledgment(ACK) indicating
that it has a go od radio connectivity and has processed the update message.
Because of the broadcast nature of the radio channel, a node can send a single
update message to inform all its neighbors about changes in its routing table;
however, each such neighbor sends an ACK to the originator node.
In addition to ACKs, the connectivity can also b e ascertained with the receipt
of any message from a neighbor (which need not b e an update message). To
ensure that connectivity with a neighbor still exists when there are no recent
transmissions of routing table updates or ACKs, perio dic update messages without
any routing table changes (null update messages) are sent to the neighbors. The
time interval between two suchnull up date messages is the
Hel loInterval
.
If a node fails to receiveanytyp e of message from a neighbor for a specied
amount of time (e.g., three or four times the HelloInterval known as the
Router-
DeadInterval
), the node must assume that connectivity with that neighbor has
been lost.
2.2 { Information Maintained at Each Node
For the purpose of routing, each no de maintains a
distance table
,a
routing table
,
a
link-cost table
and a
message retransmission list
.
The distance table of node
i
is a matrix containing, for each destination
j
and each neighbor of
i
(say
k
), the distance to
j
(
D
i
jk
) and the predecessor (
p
i
jk
)
reported by
k
.
S. Murthy and J.J. Garcia-Luna
5
Procedure Init1
when
router
i
initializes itself
do begin
set a link state table with costs of adjacent links;
N
i
;
N
i
x
j
l
i
x
<
1
;
for each
(
x
2
N
i
)
do b egin
N
i
N
[
x
;
tag
i
x
null
;
s
i
x
null
;
p
i
x
null
;
D
i
x
1
end
D
i
i
0;
s
i
i
null
;
p
i
i
null
;
tag
i
i
corr ect
for each
j
2
N
call
Init2
(
x; j
)
for each
(
n
2
N
i
)
do
add (0
;i;
0
;i
)to
LI ST
i
(
n
)
x
retransmission time;
y
hello count;
z
retransmission count;
call
Send
end
Procedure Init2
(
x; j
)
begin
D
i
jx
1
;
p
i
jx
null
;
s
i
jx
null
;
seqno
i
jx
0
end
Procedure Send
begin
for each
(
n
2
N
i
)
do begin
if
(
LI ST
i
(
n
) is not empty)
then
send messages with
LI ST
i
(
n
)to
n
empty
LI ST
i
(
n
)
end
end
Procedure Message
when
router
i
receives a message on link (
i; k
)
begin
if
(
k
62
N
i
)
do
begin
N
i
N
i
[
k
;
l
i
k
cost of new link;
if
(
k
62
N
)
begin
N
N
[
k
;
tag
i
k
null
;
D
i
k
1
;
p
i
k
null
;
s
i
k
null
;
for each
x
2
N
i
do
call
Init2
(
x; k
)
end
for each
(
i; k ; l
i
k
)
do
send
update(0
;k;D
i
k
;p
i
k
)
end
reset HelloTimer;
for each
entry (
u
k
j
;j;RD
k
j
;rp
k
j
)
j
i
6
=
j
do begin
if
(
j
62
N
)
then b egin
if
(
RD
k
j
=
1
)
then
delete entry
else b egin
N
N
[
j
;
for each
entry
x
2
N
i
call
Init2
(
x; j
)
tag
i
j
null
; call
DT
end
end
else b egin
tag
i
j
null
;
end
end
for each
entry (
u
k
j
;j;RD
k
j
;rp
k
j
) left
j
i
6
=
j
do case of
u
k
j
0: call
Update
(
j; k
)
1: call
ACK
(
j; k
)
end
call
Send
end
Procedure Create RList
(
seqno
)
begin
seqno
seqno
+1;
N eighbor Set
N
i
bitmap
[]
0;
RetransmissionT imer
x
add updates to RList
end
Procedure Delete RList
(
seqno
)
begin
set
bitmap
[
seqno
]
1;
delete
1
for all
n
2
N
i
begin
if
(
bitmap
[
seqno
]=0)
delete
0;
end
if
(
delete
= 1) delete RList[seqno]
end
Procedure Update RList
(
seqno
)
begin
reset RetransmissionTimer
send update
RList
[
seqno
];
end
Procedure Clean RList
(
seqno
)
begin
for all
entries in
RList
delete
RList
[
seqno
];
end
Procedure Connectivity
when
HelloTimer expires
begin
H elloCount
[
k
]
H elloCount
[
k
]+1;
if
(
H elloCount
[
k
]
<y
)
then
reset HelloTimer;
else begin
N
i
N
i
,
k
call
Delete RList
(
k
)
l
i
k
1
tag
i
k
null
delete column for
k
in distance table
update routing table
end
end
Procedure TimeOut
(
i; k
)
when
RetransmissionTimer expires
begin
RetransmissionCounter
RetransmissionCounter - 1;
if
(RetransmissionCounter
<
z)
call
Update RList
(
k
)
else begin
N
i
N
i
,
k
call
Delete RList
(
k
)
l
i
k
1
tag
i
k
null
delete column for
k
in distance table
update routing table
end
end
Procedure DT
when
distance table up date has to be done
begin
D
i
jk
l
i
k
+
D
k
j
;
p
i
jk
p
k
j
;
(2)
for all
neighbors
b
do begin
if
k is in the path from i to j in
the distance table through neighbor
b
then
D
i
jb
D
i
kb
+
D
k
j
;
p
i
jb
p
k
j
end
end
Figure 1: Protocol Specication