QoS Implementation and Evaluation for Mobile Ad hoc
Xuefei Li, Laurie Cuthbert
Department of Electronic Engineering, Queen Mary, University of London,
Mile End Road, London, UK
Abstract. Future mobile Ad hoc networks (MANETs) are expected to be based
on all-IP architecture and be capable of carrying the multitude of real time mul-
timedia applications such as voice, video and data. It is very necessary for
MANETs to have an efficient routing and quality of service (QoS) mechanism
to support diverse applications. In this paper, we propose a novel node-disjoint
Multipath QoS Routing protocol for supporting DiffServ (MQRD). Simulation
results show that MQRD achieves better performance in terms of packet deliv-
ery ratio and average delay.
1 Introduction
With the rising popularity and development of wireless network based on all-IP archi-
tecture and multimedia applications, potential uses of MANETs in military and civil-
ian life are attracting more and more researchers’ attentions on QoS support. QoS
provisioning in MANETs faces a number of technical challenges because of the net-
work restrictions such as dynamically and unpredictably variable topology resulting
from nodal mobility and bandwidth constraints caused by the shared wireless me-
For the current Internet there are two different models to obtain a QoS guarantee:
the Integrated Services (IntServ) [1]
and Differentiated Services (DiffServ) [2]. Int-
Serv uses the RSVP protocol [3] to carry the QoS parameters from the sender to the
receiver to make resource reservations along the path. IntServ/RSVP provides for a
rich end-to-end QoS solution, by way of end-to-end signalling, state-maintenance (for
each RSVP-flow and reservation), and admission control at each network element.
DiffServ on the other hand, does not have any end-to-end signalling mechanism and
works on a service level agreement between the provider and the user. Multiple flows
in DiffServ model are mapped to a single service level and state information about
every flow need not be maintained along the path.
IntServ-based model on per-flow resource reservation is not particularly
for MANETs because of the frequently changing topology and limited resources in
MANETs, resulting in more signalling overhead and unaffordable storage and com-
puting process for mobile nodes. DiffServ-based is a lightweight model using a rela-
tive-priority scheme to soften the hard requirements of hard QoS models like IntServ.
The service differentiation is based on per-hop behaviours (PHBs) [4], so no flow
Li X. and Cuthbert L. (2005).
QoS Implementation and Evaluation for Mobile Ad hoc Networks.
In Proceedings of the 2nd International Workshop on Ubiquitous Computing, pages 82-87
DOI: 10.5220/0002576200820087
states need to be maintained within the core of the network. Thus the model could be
a potential QoS model in MANETs.
The current existing solutions for QoS provisioning in MANETs are mainly
based on the IntServ or DiffServ model. AQOR [5] uses a reservation-oriented
method to decide admission control and allocate bandwidth for each flow. INSIGNIA
[6] employs an in-band signalling protocol rather than out-of-band signalling protocol
as RSVP to decrease reservation overhead. FQMM [7] is designed to provide QoS in
ad hoc networks by mixing the IntServ and DiffServ mechanisms. High priority ap-
plications are provided by IntServ per-flow QoS guarantee, while lower priority ap-
plications are provided with per-class differentiation based on DiffServ. SWAN [8] is
based on reservation-less approach. By avoiding signalling, it simplifies the whole
architecture and provides a differentiation between real-time and best effort in spite
of not being able to guarantee the QoS needs of each flow for the whole session due
to frequently changing topology and limited wireless bandwidth restriction.
Multipath routing allows the establishment of multiple paths between a single
source and single destination node during a single route discovery. Some multipath
routing protocols [9,10] in MANETs have been proposed to provide load balancing,
fault-tolerance and higher aggregate bandwidth as well as eliminate route discovery
latency after a link break by making use of the availability of multiple route paths.
However, these multipath routing protocols lack QoS support in the process of trans-
mission of data packets.
In this paper, we present a novel node-disjoint Multipath QoS Routing protocol
for supporting DiffServ (MQRD), which combines DiffServ and a multipath routing
protocol, Node-Disjoint Multipath Routing (NDMR) [9], for QoS provisioning in
Although NDMR provides node-disjoint multipath routing with low route overhead
in MANETs, it is only a best-effort routing approach, which is not enough to support
QoS. DiffServ is an approach for a more scalable way to achieve QoS in an IP net-
work. It could be a potential QoS model in MANETs because it acts on aggregated
flows and minimises the need for signalling. However, one of the biggest drawbacks
of DiffServ comes from the fact that the QoS provisioning happens separate from the
routing process.
2.1 Integration of NDMR and DiffServ
Both of NDMR and Diffserv operate at the network layer, so it is easy to work natu-
rally together. Although NDMR was designed without taking QoS into consideration,
it and DiffServ could be complementary techniques that can be implemented in
MANETs to support an end-to-end QoS solution. When used together, DiffServ pro-
vides the standardized QoS mechanisms and NDMR provides node-disjoint multipath
routing techniques increasing the network resource optimization and decreasing rout-
ing overhead.
2.2 QoS and Resource Management of MQRD
Effective QoS mechanism can be used to provide better service to certain flows in the
environments of limited wireless bandwidth. In MQRD this is done by either raising
the priority of a flow or limiting the priority of another flow. In order to support ser-
vice differentiation, scheduling and queue management are thought to be two impor-
tant aspects of resource management. The former is done by the scheduler which
decides the opportunities of flows for link access and the latter holds the valid packets
when necessary drops some packets from the buffer in case of network congestion.
Priority Scheduling. In MANETs, when a mobile node is receiving traffic faster
than it can transmit, the node may buffer the extra traffic until bandwidth is available.
In MQRD, priority queuing is used to build a priority scheduler. The priority sched-
uler includes two queues: a high-priority queue and a low-priority queue. The high-
priority queue must be emptied before packets are emptied from low-priority queues.
Although DiffServ has a lot of classes defined, the most essential use of DiffServ
is to provide support for the two most common applications:
(A) Voice, Video traffic. (B) Best effort data.
Let us denote the two classes as A and B. Class A applications require generally
low loss, low latency and assured bandwidth service, so packets of class A are classi-
fied as Expedited Forward (EF) traffic. Class B is classified as Best Effort (BE) traf-
fic which offers a lower priority service. Our priority scheduler (see Fig. 1) is de-
signed to transmit any available Class A packets ahead of Class B packets. On the
other hand, Class B packets are not sensitive to delay, as the application which they
service are primarily HTTP and FTP sessions.
Class A
Class B
Low Priority Queue
High Priority Queue
Fig. 1. Priority Scheduler
Queue Management. While the scheduling does play a big role in the QoS provided
by the network, it is only effective if there is sufficient queue space to hold incoming
packets. Because queues are not of infinite size, they can fill and overflow. When a
queue is full, any additional packets cannot get into the queue and will be dropped.
This is a tail drop. The issue with tail drops is that the router cannot prevent this
packet from being dropped (even if it is a high-priority packet). So, the purpose of
queue management is to make sure the queue does not fill up so that there is room for
high-priority packets.
The random early detection (RED) algorithm [11] is implemented to avoid conges-
tion before it becomes a problem. The minimum threshold specifies the number of
packets in a queue before the queue considers discarding packets. The probability of
discard increases until the queue depth reaches the maximum threshold. After a queue
depth exceeds the maximum threshold, all other packets that attempt to enter the
queue are discarded.
2.3 Load Balance and Congestion Avoidance
As mentioned before, MQRD can discover multiple node-disjoint route paths with
low routing overhead, so it can provide load balancing and higher aggregate band-
width. Load balancing function can be triggered to avoid congestion by spreading the
traffic along multiple routes when RED algorithm judges the queue depth to reach the
minimum threshold at which the queue begins to consider discarding packets. The
mobile node needs to send a Congestion Notification packet (CN) to the source of the
data packet along the reverse route path. When the source receives the CN, it distrib-
utes part of traffic to the other node-disjoint routing paths. In this way congestion and
bottleneck are avoided or alleviated.
3 Simulation Model
OPNET 8.1 Modeler [12] was used to create a simulation environment to develop and
analyze the proposed node-disjoint Multipath QoS Routing protocol for supporting
DiffServ (MQRD) and compare performances with NDMR, which does not take QoS
into account.
The random waypoint model [13] is used to model mobility. Each node starts its
journey from a random location to a random destination with a random velocity of 0-
20 m/s. Field configuration of 1000m x 1000m field with 50 nodes is used and each
node uses the IEEE 802.11[14] with a 250m transmission radius. Traffic sources with
512 byte data packets are CBR (constant bit rate). The source-destination pairs are
spread randomly over the network and the number of sources is varied to change the
offered load in the network.
In order to investigate the usage of network ability, the number of EF (Expedited
forwarding) sources with 80kbit/s (20pkt/s) bandwidth requirement is varied from 5
to 20 in intervals of 5. 20 other nodes are randomly chosen to send background BE
(Best Effort) traffic with 2pkt/s. Simulations are run for 800 simulated seconds.
4 Simulation Results
Comparing Fig. 2 and Fig. 3, we can find that the packet delivery ratio of MQRD has
better performance than that of NDMR with the increase in the number of EF sources.
In order to show clearly and compare simulation results of different type of packets,
packet delivery ratios of EF packets, BE packets and ALL packets (combination of
EF and BE packets) are depicted respectively in the two figures. Fig. 3. shows that EF
packets have higher delivery ratio than BE packets because priority scheduler is used
in MQRD. When the number of EF sources increases, NDMR drops a larger fraction
of the packets than that of MQRD. The reason is that there exists more congestion in
mobile node buffers when the number of EF sources increases.
From Fig. 4 we can see that EF packets and BE packets in NDMR have little dif-
ference in End-to-End average delay. The reason is that there is no priority policy to
deal with the incoming EF and BE packets in mobile nodes. Fig. 5 shows that EF
packets of MQRD has a much lower average delay than BE packets because priority
scheduler in MQRD makes EF packets be forwarded more quickly. With the increase
in the number of EF sources average delay of BE packets in MQRD increases more
quickly than that of EF packets. The reason is that an increase in the number of EF
sources leads to higher network load traffic. Because of the limitation of a con-
strained wireless bandwidth, BE packets that will be sent or forwarded have to stay in
buffers and wait for a longer time to get a radio channel available than EF packets in
order to avoid traffic congestions.
5 10152
Number of EF Sources
Packet Delivery Rati
EF pkts
BE pkts
ALL pkts
Number of EF Sources
Packet Delivery Rati
EF pkts
BE pkts
ALL pkts
Fig. 2. Packet Delivery Ratio of NDMR Fig. 3. Packet Delivery Ratio of MQRD
Number of EF Sources
Average Delay (ms
EF pkts
BE pkts
ALL pkts
Number of EF Sources
Average Delay (ms
EF pkts
BE pkts
ALL pkts
Fig. 4. Average Delay of NDMR Fig. 5. Average Delay of MQRD
5 Conclusions
In this paper, we have introduced a practical QoS provisioning which makes DiffServ
over node-disjoint multipath routing protocol for MANETs to overcome the short-
comings of the best-effort model. We also present a solution about a reliable multi-
path routing and resource management for QoS issues of real-time multimedia appli-
cations in ad hoc networks. The performance evaluation and comparison between
NDMR and MQRD are studied by extensive simulations using OPNET Modeler.
Simulation results show that MQRD achieves better performance than NDMR by
providing end-to-end QoS support in MANETs. We can conclude that MQRD has a
good potential to serve as a QoS model to provide real-time multimedia applications
under the dynamically changing environment of ad hoc networks.
1. R. BBraden, D. Clark, and S. Shenker, “Integrated Services in the Internet Architecture –
an Overview”, IETF RFC1663, June 1994.
2. S. Blake, “An Architecture for Differentiated Services”, IETF RFC2475, December 1998.
3. R. Braden, L. Zhang, S.Berson, S. Herzog, and S. Jamin, “Resource reservation Protocol
(RSVP) – Version 1 Functional Specification”, RFC2205, Sept. 1997.
4. D. Black, S. Brim, B. Carpenter, “Per Hop Behavior Identification Codes”, June 2001
5. Q. Xue and A.Ganz. Ad hoc QoS on-demand routing (AQOR) in mobile ad hoc networks.
Journal of parallel and Distributed Computing, 154-165, 20003
6. X. Zhang S.B. Lee, A. Gahng-Seop and A.T. Campbell, INSIGNIA: An IP-Based Quality
of Service Framework for Mobile Ad hoc Networks. Journal of Parallel and distributed
Computing, 60(4), 374-406, April 2000
7. A. Lo H.Xiao, W.K.G..Seah and KC Chua. A Flexible Quality of Service Model for Mo-
bile Ad hoc networks. In IEEE Vehicular Technology Conference Fall 2000, pages 445-
449, May 2000
8. A. Veres G.Ahn, A.T. Campbell and L.Sun. SWAN: Service Differentiation in Stateless
Wirelass Ad hoc network. In Conference on Computer Communications, Jun 2002
9. X. Li and L. Cuthbert, Stable Node-Disjoint Multipath Routing with Low Overhead in
Mobile Ad Hoc Networks, In Proceedings of the IEEE MASCOTS’04, Netherland, Oct.
10. S.J.Lee and M.Gerla, Split Multipath Routing with Maximally Disjoint Paths in Ad Hoc
Networks, In Proceedings of the IEEE ICC, pages 3201-3205, 2001.
11. S. Floyd and V. Jacobson, Random Early Detection Gateways for Congestion Avoidance
12. OPNET Technologies, Inc. http://www.opnet.com/
13. 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 MobiCom’98, Oc-
tober 25-30, 1998, Dallas, Texas, USA.
14. IEEE Standards Department, Wireless LAN medium access control (MAC) and physical layer
(PHY) specifications, IEEE standard 802.11, 1997