An Opportunistic Transmission Mechanism

for Seamless Video Streaming

in the MIMO-capable Wireless Ad Hoc Network

Hayoung Oh Sanghyun Ahn*, Yang Cheng

Soongsil University University of Seoul
School of Electronic Engineering School of Computer Science
Seoul, Korea Seoul, Korea
hyoh@ssu.ac.kr *ahn@uos.ac.kr,cy00002009@gmail.com

network may carry more traffic and easily get congested. The
Abstract The Multiple Inputs Multiple Output (MIMO) antenna
congestion problem affects the performance of the entire
technique provides two transmission modes, MUX and DIV. The
network in terms of delay and throughput.
MUX mode increases data rate and, on the other hand, the DIV mode
reliability or transmission range. In this paper, we propose an The multipath scheme can improve network reliability in
opportunistic transmission mechanism that adaptively changes the case of link disconnection of an active route. The multipath
MIMO transmission mode to provide reliability for the seamless routing approach is much more robust than the single path
video streaming service. Also, our mechanism utilizes an extra detour routing scheme and requires less cost for the alternate route
path for resolving congestion. We aim to satisfy various QoS discovery. The Ad hoc On-Demand Multipath Distance Vector
requirements of multi-priority video data packets by assigning (AOMDV) [2], an extension of the Ad hoc On-Demand
different paths and MIMO modes to different priority packets. Distance Vector (AODV) [3], was proposed for computing
Thanks to the low overhead of a single detour path and the
multiple loop-free and link-disjoint paths. Lee et al. [4]
opportunistic MIMO transmission mode change, our proposed
mechanism becomes a proper solution for the seamless video
proposed the Split Multipath (SMR) protocol based on DSR
streaming service. By carrying out extensive simulations, we show that can find an alternate route maximally disjoint from the
that our mechanism improves the network throughput.1 shortest delay route. Jenn-Yue Teo et al. [5] proposed the
Interference-Minimized Multipath Routing (I2MR) protocol to
increase throughput by discovering zone-disjoint paths for load
Keywords MIMO, Wireless Ad Hoc Network, Seamless Video balancing, requiring minimal localization support. In [6], the
Streaming, Multipath Routing authors propose an on-demand multiple path routing protocol
based on the Dynamic Source Routing (DSR) [7] for mobility.
I. INTRODUCTION In [8], the authors propose an on-demand multipath routing
protocol for multi-hop wireless networks that can find spatially
In recent years, the Multiple Input Multiple Output
disjoint paths without location information. [9] shows that
(MIMO) becomes one of the popular technologies for wireless
finding multiple uncorrelated node-disjoint paths between a
communications. The MIMO antenna technique provides two
given pair of nodes in an ad hoc network is the same as finding
options of operation: (a) to transmit different data
a chord-less cycle in a graph that contains source and
simultaneously by using multi-antennas for the increase of data
destination nodes. Wang et al. [10] proposed an approach via
rate, called the Spatial Multiplexing (MUX), and (b) to transmit
geographic routing to restrict the route request flooding and
the same data simultaneously by using multi-antennas for the
guarantee no collision between a pair of nodes by making them
increase of the reliability or the transmission range, called the
apart from each other by the transmission range.
Spatial Diversity (DIV) [1].
However, previous multipath routing protocols have
If the MIMO MUX mode is used for packet transmission,
message and/or storage overhead proportional to the number of
the network throughput can be improved. However, it is also
nodes and the number of transmission failures. And they did
possible that a node may get overwhelmed (congested) by the
not consider timely throughput for seamless video streaming
packets from its neighbors using the MUX mode. If we
and the congestion problem.
consider the multi-source network environment, this congestion
problem can get even severe. Since lots of connections are set Therefore, in this paper, we propose an opportunistic
up in a network, wireless links located at the center of the transmission mechanism for the seamless video streaming in
the MIMO-capable wireless ad hoc network. Our mechanism
utilizes an extra detour path and the MIMO mode change to
*
Corresponding Author: Sanghyun Ahn, ahn@uos.ac.kr mitigate congestion by dispersing traffic. Also, we support
1
This research was supported by the MSIP (Ministry of Science, ICT different QoS requirements of multi-priority packets by
and Future Planning), Korea, under the ITRC (Information dedicating different paths and MIMO modes to different
Technology Research Center) support program (NIPA-2014-H0301- priorities.
14-1015) supervised by the NIPA (National ICT Industry Promotion
Agency).

978-1-4799-3494-2/14/$31.00 2014 IEEE 443 ICUFN 2014

 The rest of the paper is organized as follows. The related A routing algorithm with QoS provisioning is presented in
work is briefly covered in Section II and the detailed [13] to exploit the multiplexing gain and the interference
description of the proposed mechanism is given in Section III. cancelation property of MIMO antennas.
Section IV presents the simulation results. Finally, we conclude
the paper in Section V. Compared with previous work, to the best of our
knowledge, this is the first work putting forward an
opportunistic on-demand detour-path routing mechanism and,
II. RELATED WORK at the same time, considering the effect of different QoS
The multipath routing can reduce the need for route updates, requirements of video streaming in the MIMO-capable wireless
balance the traffic load, increase the data transfer rate in a ad hoc network.
wireless network and improve the utilization of the limited
resources of a wireless network. III. PROPOSED SCHEME
In [4], the authors propose the Split Multipath (SMR) In this section, we define three types of transmission
protocol based on DSR that can find an alternate route policies and the system model, and describe our proposed
maximally disjoint from the shortest delay route. S. J. Lee and multipath routing mechanism in detail.
M. Gerla [5] proposed the Interference-Minimized Multipath
Routing (I2MR) protocol to increase throughput by discovering A. Transmission Policies
zone-disjoint paths for load balancing with requiring minimal We define three transmission policies as follows:
localization support.
P1: A shortest path with MUX. This policy adopts a
A. Nasipuri et al. [6] also focused on DSR and showed how shortest path using MUX. The shortest path is
the intelligent use of multipath techniques can reduce the defined as the minimum latency path constructed by
frequency of query floods. They developed an analytic the DSR routing protocol. MUX is the MIMO
modeling framework to determine the relative frequency of transmission mode used to increase the data rate.
query floods for various techniques.
P2: A detour path with MUX. This policy adopts a
In [8], the authors propose an on-demand multipath routing detour path using MUX. The detour path is the
protocol for the multi-hop wireless network that finds spatially shortest among the paths that do not overlap with the
disjoint paths physically distant from each other without the shortest path the most. This policy resolves
need of location information. [9] shows the impact of congestion and has a higher priority than the policy
interference patterns on the throughput of a wireless network. P3.
In [9], multiple uncorrelated node-disjoint paths between a
given pair of nodes are found based on a scheduling strategy by P3: A shortest path with DIV. This policy adopts a
taking into account global information as opposed to the local shortest path using DIV. DIV is the MIMO
information considered in the link and node based strategies. transmission mode used to increase the transmission
radius. When congestion happens, this policy makes
Wang et al. [10] proposed an approach via geographic the congested node bypassed by packets, resulting in
routing to restrict the route request flooding to the neighbors of less traffic at the congested node. This policy
nodes and guarantee no collision between the routes where requires more processing than P2, so P2 is more
each pair of nodes has to be apart from each other by the preferred than P3 in resolving congestion.
transmission range.
Recently, MIMO-based routing protocols have been B. System Model
proposed to maximize the performance gain in wireless For the opportunistic transmission mechanism, we
networks. They can be categorized into multipath and single- introduce two types of queue thresholds. The current node
path routing protocols. K. Sundaresan et al. presented a queue size is measured against the queue thresholds and the
representative MIMO routing protocol, MIR [11], based on current node queue size is defined as follows:
DSR. In MIR, when nodes are close to each other, the MUX
mode is used and the DIV mode is used only for disconnected m

links. Therefore, the MIR scheme may introduce congestion at

i
ni
=1 (1)
=
some specific nodes since nodes tend to utilize the MUX mode MAX (c )
as much as possible.
where
Y.-S. Chen et al. [12] proposed a bow-based multipath
routing protocol to enhance the MIR protocol by continuously : the threshold value
checking the link states of the MIMO ad hoc network. i : the packet index in the node queue (the ith packet)
However, to maintain and use multiple paths, this work needs
the additional overhead of periodic beacon messages and can m : the total number of packets
operate only in the time division multiple access (TDMA) ni : the number of bytes of the ith packet
system.
MAX(C) : the maximum capacity of the node queue

444
 Two types of queue thresholds, THm-d and THd-m (THm-d > C. Congestion Control Algorithm with Video Streaming
THd-m), are defined. The threshold THm-d is used to indicate As described in the previous subsection, upon receiving an
the node to reduce the incoming traffic to avoid congestion. If mRERR message, the previous-hop node changes its
is greater than THm-d, the node recognizes that it is transmission policy from P1 to P2 or P3 based on the network
experiencing congestion and figures out the previous-hop node condition. At first, the previous-hop node of the congested
which has sent packets the most and sends a control message, node changes its transmission policy to P2. The previous-hop
called a modified Route Error (mRERR) message, requesting node may use its pre-determined detour path or initiate the
for a transmission policy change to that node. Upon receiving route discovery. This detour path must be disjoint from the
the message, the previous-hop node switches its transmission formerly used (or original) path and, if there are more than one
policy from P1 to P2 or P3 based on the network condition. disjoint path, the shortest one is chosen. Only the paths whose
The threshold THd-m is used to indicate the node to increase hop counts are not larger than that of the original path plus 3
the traffic for higher throughput. If is less than THd-m, the can be the candidates for a detour path (for the reason of using
congested node sends an mRERR message to the previous-hop 3, refer to Section IV). If the previous-hop node cannot find a
node requesting the node for changing the transmission policy detour path, it changes its transmission policy to P3.
from P3 to P1.
If the queue length of the congested node becomes less than
the threshold THd-m, the congested node sends an mRERR
message to the previous-hop node again. If the transmission
policy of the previous-hop node is P3, it changes its
transmission policy to P1 in order to improve the network
throughput.

D. Video Streaming
We assume that the video traffic is categorized into three
types of video frames; I-frames, P-frames and B-frames. I-
frames are known to be the most important frames which do
not need the other ones to get decoded. On the other hand, P-
and B-frames should refer to the previous and the following I-
frames to get decoded. Hence, the dependency on I-frames is
much higher than that on P- and B-frames. Thus, assigning
Figure 1. Scenario with congested nodes shorter paths to I-frames is vital. Upon congestion, P- and B-
frames are transferred over to a detour path because they are
Figure 1 illustrates the operation of our proposed the second and third levels of importance from the perspective
mechanism. Node pairs (S1, D1) and (S2, D2) are two pairs of of video stream reconstruction.
sources and destinations. Initially, each node sends packets by
using MUX to improve the overall network throughput and IV. PERFORMANCE EVALUATION
periodically checks the queue size of its own. When the queue
We carried out simulations to evaluate the performance of
of node C temporarily exceeds the threshold THm-d, it is
our proposed mechanism by using the Qualnet simulator [14].
considered as congestion. Then, node C finds the previous-hop
We have modified DSR such that each source uses its shortest
node A which has sent packets the most by checking the DSR
alternate route as its detour path and we call the scheme the
header and then sends an mRERR message to A to request for
modified DSR (mDSR). Our mechanism is compared with
using the P2 or P3 transmission policy. And node A tries to
mDSR and the MIMO policy only scheme based on MIR. The
find the most different path from the former one as a detour
simulation parameters are shown in Table 1. The network size
path, so node A finds the path A-I-E-F-G-D1 and sends low
priority packets such as B- and P-frames by using this path, is set to 1,500 m 1,500 m, and the simulation time to 300
and sends high priority packets such as I-frames along the seconds. The video encoding method is MPEG and the MPEG
former path. traffic is generated with the pattern of IBBPBBPBBPBB. 50
nodes are randomly placed in the simulation area. The radio
Now, let us assume that node H also gets congested. If the propagation range and the channel capacity of the DIV mode
previous-hop node C after receiving the mRERR message are chosen to be 750 m and 6 Mbps, respectively. The channel
from node H cannot find a detour path to the destination, C capacity of the MUX mode is 9 Mbps which is higher than
changes its transmission policy to P3 (i.e., sends packets by DIV, and the radio propagation range is 400m. We adopt the
using DIV) and sends I-, B- and P-frames by using the path C- IEEE 802.11g.
J-D2. After congestion is resolved, the congested node H
sends an mRERR message again to the previous-hop node to
request for the transmission policy change from P3 to the
original P1 in order to improve the overall throughput of the
network.

445
 Table 1. Simulation parameters with 50 nodes. This figure also indicates that the throughput of
the proposed mechanism is higher than the others.
Parameter Value
Simulation area Figure 6 is the graph showing the throughput for the hop
1500 m 1500 m
count difference between the detour path and the shortest path
Traffic Model VBR (i.e., (the hop count of the detour path) (the hop count of the
Simulation Time 300 s shortest path)). When the hop count difference is 3, the MAC
MAC Protocol 802.11g layer interference between the detour path and the shortest path
Video Encoding Method MPEG is small enough that the performance becomes the best. On the
other hand, if the hop count difference is greater than 3, the
Source-Destination pairs 2 detour path performs poorly. Hence, in the latter case, we adopt
Number of Antennas 2 to use P3 instead of the detour path for congestion resolution.
Queue Check Time 0.5 s
Figure 7 shows the throughput for various numbers of
80% source and destination pairs. More pairs mean higher chance of
50% congestion and of interference in the middle of the shortest
Speed(MUX) 9 Mbps path. Therefore, the throughputs of the three schemes decrease
Speed(DIV) 6 Mbps as the number of pairs increases. However, the proposed
scheme still gives higher throughput than the other schemes.

Figure 2 shows a simple network topology with two flows

realized with the Qualnet.

Figure 3. Traffic changes in terms of the queue size at the

congested node 5

Figure 2. A simple topology with two flows

Figure 3 shows the traffic change in terms of the queue size

at the congested node 5 in Figure 2. The current queue size of
the node is obtained from Eq. (1). We can see that our
mechanism has smaller queue size almost all the time except
for the initial stage. Because our scheme utilizes a detour path
to transmit P- and B-frames, load balancing effect can be
achieved, resulting in smaller queue size. The reason for the
high fluctuation of the MIMO policy only scheme is frequent
changes of the transmission policy between P1 and P3.
Figure 4 shows the throughput for the simple topology in
Figure 4. Throughput of the simple topology in Figure 3
Figure 2. Throughput is the amount of data transferred from
sources to destinations during the given simulation time. This
figure obviously shows that our proposed mechanism gives
higher throughput than the others thanks to the detour path
policy. Figure 5 shows the throughput for a random topology

446
 REFERENCES
[1] D. Gesbert, M. Shafi, D. Shiu, P. J. Smith and A. Naguib,"From Theory
to Practice: An Overview of MIMO Space-Time Coded Wireless
Systems", IEEE JSAC, vol. 21, is. 3, pp. 281-302, 2003.
[2] M. K. Marina and S. R. Das, "On-demand Multipath Distance Vector
Routing in Ad Hoc Networks", 9th ICNP, 2001.
[3] C. Perkins, E. Belding-Royer and S. Das, "Ad hoc On-Demand Distance
Vector (AODV) Routing", IETF RFC 3561, 2003.
[4] S. J. Lee and M. Gerla. Split Multipath Routing with Maximally
Disjoint Paths in Ad Hoc Networks, IEEE International Conference on
Communications (ICC), pp. 3201-3205, 2001.
[5] J. Y. Teo, Y. Ha, and C. K. Tham, Interference-minimized multipath
routing with congestion control in wireless sensor networks for high-rate
Figure 5. Throughput of a random topology with 50 nodes streaming, IEEE Transactions on Mobile Computing, vol. 7, no. 9, pp.
11241137, 2008.
[6] A. Nasipuri and S. R. Das, On-demand Multipath Routing for Mobile
Ad Hoc Networks, 8th International Conference on Computer
Communications and Networks, 1997
[7] D. Johnson, Y. Hu and D. Maltz, "The Dynamic Source Routing
Protocol (DSR) for Mobile Ad Hoc Networks for IPv4", IETF RFC
4728, 2007.
[8] J. J. Galvez, P. M. Ruiz and A. F. G. Skarmeta, Multipath routing with
spatial separation in wireless multi-hop networks without location
information, Computer Networks, vol. 55, no. 3, pp. 583599, 2011.
[9] A. B. Mohanoor, S. Radhakrishnan and V. Sarangan, Interference
Aware Multi-path Routing in Wireless Networks, Proceedings of 5th
IEEE In- ternational Conference on Mobile Ad Hoc and Sensor Systems
(MASS), October 2008.
[10] Z. Wang, E. Bulut and B. K. Szymanski, Energy efficient collision
aware multipath routing for wireless sensor networks, IEEE
Figure 6. Throughput vs. Hop count difference International Conference on Communications (ICC), 2009.
[11] K. Sundaresan and R. Sivakumar, "Routing in Ad-hoc Networks with
MIMO Links", 13th ICNP, 2005.
[12] Y.-S.Chen, C.-S. Hsu and P.-T. Hsieh, "An Efficient Bow-based On-
Demand QoS Routing Protocol for MIMO Ad Hoc Networks",
Computer Communications, vol. 32, is. 15, pp. 1613-1630, 2009.
[13] A. Gkelias, F. Boccardi, C. H. Liu and K. K. Leung, "MIMO Routing
with QoS Provisioning", 3th ISWPC, 2008.
[14] QualNet. Available from: http://web.scalable networks.com/
content/qualnet.

Figure 7. Throughput vs. Number of source and

destination pairs

V. CONCLUSION
In this paper, we proposed an opportunistic transmission
mechanism for the seamless video streaming in the MIMO-
capable wireless ad hoc network. Due to the low overhead of
using a single detour path and the opportunistic MIMO mode
change, our proposed mechanism is a proper solution for the
seamless video streaming. By performing extensive
simulations, we showed that our proposed mechanism
improves network performance than mDSR and the MIMO
policy only scheme. As a future work, we will figure out the
optimal node queue thresholds for the opportunistic MIMO
mode change.

447