Obtained results show characteristic differences in
performance between considered routing protocols, which
are the consequence of various mechanisms on which
protocols are based. Although we carried out simulations
with 10, 20 and 30 sources, only results with 20 sources
are presented in this paper.
As in [1], “on-demand” routing protocols, AODV and
DSR, achieve high values of PDR, which means they are
efficient protocols from the point of delivering packets to
their destination (Fig. 1). For AODV and DSR protocols,
PDR is independent of mobility and number of sources,
while DSDV has approximately the same PDR under low
mobility. As shown in Table 1 and Table 2, AODV and
DSR protocols deliver over 90% of packets for all
considered values of pause time and maximum movement
speed. Since DSDV protocol uses a “table driven”
approach of maintaining routing information, it isn’t
adaptive to the route changes that occur under high
mobility as AODV and DSR protocols are. That is why it
delivers less data packets, which is also shown in Table 1
and Table 2. Therefore, DSDV protocol is not suitable for
MANETs, because it is slow.
Results obtained by running simulations with a
changeable sending rate confirm the previous conclusion,
but also show that all three routing protocols don’t
perform well when network load increases. As shown in
Table 3, when the network load is the largest routing
protocols deliver barely 50% of packets.
However, in all considered cases, regardless of
mobility, source number or network load, DSR protocol
generates significantly less routing load than AODV and
DSDV protocols as Fig. 2 shows. In high network load
cases, nodes using considered routing protocols send more
packets, thereby sending a larger number of routing
packets. On the other hand, in high mobility cases, link
failures happen very often. Link failures initiate route
discoveries in AODV, since nodes have only one route per
destination in their routing table. Thus, the frequency of
route discovery in AODV is directly proportional to the
contribution to AODV’s routing overhead comes from
RREQ packets. On the other hand, reaction of DSR to link
failures in comparison is mild and causes route discoveries
less often. The reason is plenty of cached routes at each
node and prolongation of route discovery until all cached
routes fail. That is the reason why RREQ packets don’t
contribute so much to DSR’s routing overhead. A large
contribution in DSR comes from RREP packets.
Analyzing average end to end delay, we come to the
conclusion that DSR routing protocol outperforms AODV
and DSDV protocols (Fig. 3), unlike the results obtained
in [1], where AODV protocol has the best performances.
As said previously, for any network topology change,
nodes that use AODV protocol have to send RREQ
packets. In other words, a route discovery process has to
be activated, because AODV is a routing protocol that has
no available route when needed. Because of inefficient
route maintenance, average end to end delay is the largest
for AODV. On the other hand, DSDV protocol proactively
holds routes to all destinations in its table, regardless of
topology changes. However, DSR protocol has the best
performances, because it doesn’t depend on periodical
activities, and it uses source routing and route caching, but
also maintains multiple routes per destination. It excels
especially in low mobility scenarios, which means that in
cases when network topology is stable, routes are not stale
and that results in the best performances under
consideration.
When a network contains a small number of sources or
node’s sending rate is low, AODV and DSDV protocols
have a similar average end to end delay as DSR, especially
when node mobility is low. In that case, the network is
less loaded. However, with source number or sending rate
increasing, network load is increasing, and average end to
end delay for all three protocols, especially AODV and
DSDV, becomes larger.
Obtained results show characteristic differences inperformance between considered routing protocols, whichare the consequence of various mechanisms on whichprotocols are based. Although we carried out simulationswith 10, 20 and 30 sources, only results with 20 sourcesare presented in this paper.As in [1], “on-demand” routing protocols, AODV andDSR, achieve high values of PDR, which means they areefficient protocols from the point of delivering packets totheir destination (Fig. 1). For AODV and DSR protocols,PDR is independent of mobility and number of sources,while DSDV has approximately the same PDR under lowmobility. As shown in Table 1 and Table 2, AODV andDSR protocols deliver over 90% of packets for allconsidered values of pause time and maximum movementspeed. Since DSDV protocol uses a “table driven”approach of maintaining routing information, it isn’tadaptive to the route changes that occur under highmobility as AODV and DSR protocols are. That is why itdelivers less data packets, which is also shown in Table 1and Table 2. Therefore, DSDV protocol is not suitable forMANETs, because it is slow.Results obtained by running simulations with achangeable sending rate confirm the previous conclusion,but also show that all three routing protocols don’tperform well when network load increases. As shown inTable 3, when the network load is the largest routingprotocols deliver barely 50% of packets.However, in all considered cases, regardless of
mobility, source number or network load, DSR protocol
generates significantly less routing load than AODV and
DSDV protocols as Fig. 2 shows. In high network load
cases, nodes using considered routing protocols send more
packets, thereby sending a larger number of routing
packets. On the other hand, in high mobility cases, link
failures happen very often. Link failures initiate route
discoveries in AODV, since nodes have only one route per
destination in their routing table. Thus, the frequency of
route discovery in AODV is directly proportional to the
contribution to AODV’s routing overhead comes from
RREQ packets. On the other hand, reaction of DSR to link
failures in comparison is mild and causes route discoveries
less often. The reason is plenty of cached routes at each
node and prolongation of route discovery until all cached
routes fail. That is the reason why RREQ packets don’t
contribute so much to DSR’s routing overhead. A large
contribution in DSR comes from RREP packets.
Analyzing average end to end delay, we come to the
conclusion that DSR routing protocol outperforms AODV
and DSDV protocols (Fig. 3), unlike the results obtained
in [1], where AODV protocol has the best performances.
As said previously, for any network topology change,
nodes that use AODV protocol have to send RREQ
packets. In other words, a route discovery process has to
be activated, because AODV is a routing protocol that has
no available route when needed. Because of inefficient
route maintenance, average end to end delay is the largest
for AODV. On the other hand, DSDV protocol proactively
holds routes to all destinations in its table, regardless of
topology changes. However, DSR protocol has the best
performances, because it doesn’t depend on periodical
activities, and it uses source routing and route caching, but
also maintains multiple routes per destination. It excels
especially in low mobility scenarios, which means that in
cases when network topology is stable, routes are not stale
and that results in the best performances under
consideration.
When a network contains a small number of sources or
node’s sending rate is low, AODV and DSDV protocols
have a similar average end to end delay as DSR, especially
when node mobility is low. In that case, the network is
less loaded. However, with source number or sending rate
increasing, network load is increasing, and average end to
end delay for all three protocols, especially AODV and
DSDV, becomes larger.
การแปล กรุณารอสักครู่..