Congestion control based on sliding mode control and scheduling with prioritized queuing for wireless networks

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 03 Special Issue: 07 |May-2014, Available @ http://www.ijret.org 840
CONGESTION CONTROL BASED ON SLIDING MODE CONTROL
AND SCHEDULING WITH PRIORITIZED QUEUING FOR WIRELESS
NETWORKS
R.Keerthana1
, S.Lavanya2
1
PG Scholar, Dept. of CSE, Sona College of Technology, Salem – 636005, TN, India
2
Asst. Professor, Dept. of CSE, Sona College of Technology, – 636005, TN, India
Abstract
The application of sliding mode control in wireless network is used for considering the joint congestion control and scheduling.
Incorporating dual decomposition enables the joint congestion control and scheduling problem to be considered separately. The
communication among these two sub problems is given by the lagrangian price value. Based on the utility optimization of network
the congestion control performance for queues can be improved using sliding mode based congestion controller. Buffers on
network devices are managed with various queuing techniques. Properly managed queues can minimize dropped packets and
network congestion, as well as improve network performance. Thus, the queuing structure of CBWFQ (Class Based Weighted
Fair Queuing) and strict Priority is combined so that the packets are classified into different classes and priorities are given to
these classes, where the higher priority are given to delay sensitive packets and are scheduled effectively.
Keywords: Sliding mode control, Joint Congestion control and Scheduling, Network Utility Maximization, Dual
Decomposition, Class Based Weighted Fair Queuing, Priority Queue, Lagrangian price.
---------------------------------------------------------------------***--------------------------------------------------------------------
1. INTRODUCTION
In recent years, there has been a progressive growth in the
field of wireless networks. In Wireless networks, bandwidth
is one of the major constraints. Some of the factors such as,
mobility, random changes in connectivity, fluctuations in
channel and interference due to neighboring nodes etc, leads
to higher rate of packet loss in a wireless network than that
of a wired network where, the packet loss occurs mainly due
to congestion in the network.
The proper congestion control mechanism is to be
incorporated in order to improve the performance of any
network. So, the congestion control is required to control the
rate of various traffic sources which inject the traffic into the
network [18]. The systematic approach to analyze and
design a system, predict system response to some input and
approaches to assess system stability has been provided by
some control theory. In wireless networks congestion
control and scheduling are the key features that are to be
considered. The joint congestion control and scheduling in
wireless network can be represented by NUM problem
which calculates the sum of link prices [1]-[3].
Sliding Mode (SM) control theory [18] is a powerful tool
that can be used for congestion control. This SM controller
can be applied to wireless networks by representing this
joint congestion control and scheduling as NUM problem.
The decomposition theory applied to the NUM problem
facilitates us to apply the SM control technique to
congestion control in ad hoc networks.
The rest of this paper is organized as follows. Section II,
describes the system model, followed by the formulation
and solution to the NUM problem, Section III extends the
solution to include queuing structures. Section IV shows
simulation results and section V concludes this paper.
2. System Model
2.1 Queuing Structure
The commonly used queuing structures are: 1) per-
destination queues 2) per-link queues. For the networks that
has per-destination queue, a separate queue is maintained for
every node for each flow. The number of queues per node is
equal to the number of nodes in the network. Overhead of
such networks would be unbearable. Whereas, in per-link
queuing networks the number of queues is based on the
number of nodes in the next-hop that is every node needs to
maintain a queue for every out going link attached with that
node.
2.2 Joint Congestion Control and Scheduling
In a network the flow of information is accomplished
through the interaction among different design layers in
order to support transfer of information. In wireless
networks, this interaction among the layers can be given by
cross-layer design, where a number of parameters are jointly
controlled. Furthermore, state information associated with a
specific layer becomes available across layers [10].
Congestion control, routing and scheduling are implemented
independently at different layers in the layered structure of

_______________________________________________________________________________________
the networks. However, in wireless networks, Congestion
control and scheduling would be jointly designed to achieve
high end-to-end throughput and efficient resource utilization
[14].
2.3 Network Utility Maximization
The joint congestion control and scheduling in wireless
networks is challenging due to unreliability, time varying
channel and interference among wireless channels.
Therefore the Network Utility Maximization (NUM)
problem can be used to formulate the joint congestion
control and scheduling [3].
Let us assume that a network has a set of resources as L and
a set of users as I. Let Cl denote the finite capacity of
resource l ∈L. Each user i ∈ I has a fixed route 𝑟𝑖 in which
each user i’s packet utilize the set of resources. 𝑥𝑖(𝑡) is the
sending rate of source. In general, consider zero-one matrix
A which is defined, where 𝐴𝑖,𝑙 = 1 if l ∈𝑟𝑖 and 𝐴𝑖,𝑙 = 0
otherwise. When its rate is 𝑥𝑖 user i receive utility𝑈𝑖(𝑥𝑖).
The utility functions of the users are used to select the
desired rate allocation among the users. The utility𝑈𝑖(𝑥𝑖) is
an increasing, strictly concave and continuously
differentiable function.
max 𝛴 𝑈𝑖(𝑥𝑖) (1)
Such that,
AT
x ≤ C (i)
xi,j − Fi,j
0
+ Fi,j
in
(ii)
Where, C = 𝐶𝑙 ∈ L is the capacity constraint which states
that the sum of the rates of all users utilizing resource should
not exceed its capacity 𝐶𝑙 .
The joint congestion control and scheduling representation
using NUM problem [3],[15] can be given as,
max Σ Ui(xi) − λi,ji,j Є L ( xi,j − Fi,j
0
+ Fi,j
in
) (2)
2.4 Dual Decomposition
Dual decomposition decomposes the original large problem
into distributive sub problems. This is mainly based on
decomposing the Lagrangian dual problem.
This method of decomposition corresponds to a resource
allocation with pricing. The original problem sets the price
for the resources to each sub problem, depending on which
it has to decide the amount of resources to be used.
From the NUM problem the congestion problem and
scheduling problem can be decomposed separately as
follows
max Σ Ui(xi) − λi,ji,j Є L ( xi,j) + max λi,ji,j Є L ( Fi,j
0
− Fi,j
in
)
(3)
2.5 Distributed Hop-By-Hop Algorithm
Necessity for full utilization of the potential capacity of the
network arises due to the scarcity of the wireless spectrum.
One approach to improve the capacity of a wireless network
is to use multi-hop instead of traditional single-hop
communication [14]. A distributed hop-by-hop algorithm is
developed for congestion control. The congestion controller
at the source reacts based on the sum of the congestion
prices at each node. In other words, each node adds its
current congestion cost to that it received from a
downstream node, and passes this information toward the
upstream node. The source will ultimately receive the sum
of all price information from the corresponding nodes and
use the information for controlling rates.
2.6 Sliding Mode Based Control
A sliding surface is constructed for switching surfaces so
that the system restricted to the switching surface produces a
desired behavior. For convenience only linear switching
surfaces of the form 𝑆𝑥 (t) = 0 are considered in practice.
For sliding mode controller switched feedback gains which
drive the state trajectory to the sliding surface and maintain
it there. For the existence of a sliding mode on the switching
surface, the state velocity vectors should be directed towards
the surface, i.e., the system must be stable to the switching
surface. Therefore a Lyapunov function can be used for
maintaining the stability. The behavior of the system on the
sliding surface is given as
L x, λ, v = f x + λii fi x + vjj hj(x) (4)
Where, 𝜆𝑖 is Lagrangian price [1], this lagrangian price can
be updated according to
λi,j t = [ yi,j X, t xi,j − Fi,j
0
+ Fi,j
in
] (5)
∀i, j ∈ L
This lagrangian price value is proportional to the queuing
length (or) delay in the node [2].
Each of the input flow 𝐹𝑖,𝑗
𝑖𝑛
and output flow 𝐹𝑖,𝑗
0
is obtained
by estimating some parameters like arrival rate, transmission
rate, and receive rate. The sliding mode controller is
designed to adjust these parameters for stable behavior of
the system, which in turn provides the feedback to control
the delay value by maintaining the flow rate to a stable value
by which congestion can be controlled [2].
Since the scheduling problem is given as
λi,j ti,j Є L ( Fi,j
0
(t) − Fi,j
in
t ) (6)
It is seen that the scheduling also depends in some way on
𝜆𝑖,𝑗 (𝑡) value. Hence by changing this value the packets
would be scheduled properly.

_______________________________________________________________________________________
2.7 Design of Sliding Mode Control Algorithm
Each user i adjusts its rate according to the following
differential equation.
d
dt
xi t = ki [ ωi − xi(t) pl[ xi t ] ]i∈IlЄri
(7)
Where,, ki and ωi are positive constants, ki is the gain
parameter, ωi shows the user’s willingness to pay per unit
time. P(t) is an increasing function of the aggregate rate of
the users going through it, and it can also be seen as the
packet loss function.
The simplified dynamic model is
r(t) = k (ω − r(t) p(t)) (8)
The dynamic buffer length at bottleneck is that
q(t) = r(t) –C (9)
Where, q(t) is the instantaneous queue length in buffer, C is
link capacity.
Let x1(t) =q t − qd and Let x2(t) =r(t) –C
This can be also given as,
x1 t = x2(t) (10)
x2(t) =k [ ω − x2 t + C ] p(t) (11)
Where qd is the reference queue length
The queue length q(t) and p(t) is the marker probability. The
queue length at congested routers is compared with the
reference value 𝑞 𝑑 and the feedback is provided. Then high
link utilization and low delay is maintained in the system.
3. PROPOSED SYSTEM
The purpose of congestion control is to maintain and control
the flow of packets that passes through the interface based
on the individual priority bits assigned to each packets. This
difficulty is overcome by effective classification and
scheduling of packets and each are assigned to the newly
created queue. There are four types of queuing protocols
which helps in valuable congestion management by creating
a different number of queues and different traffic
management by varying the sequence in which the packets
are transformed.
A network is congested when packets sending rate is faster
than the transmission rate of the interface. By utilizing the
congestion management schema, the packets are queued
until the interface gets free and they are allowed to transfer
from the corresponding interface based on the priority and
the queuing mechanism assigned to that specific interface.
The transmission among each queues are being done by the
router which determines the order of packets to be
transmitted and services that are provided by each queue.
The queuing structure which combines the feature of strict
PQ (Priority Queue) and CBWFQ (Class Based Weighted
Fair Queue) has been proposed.
CBWFQ defines classes with weights but does not provide
strict priority. The traffic classes defined by CBWFQ are
assigned with different characteristics. For example, the
characteristic of a class can be the minimum bandwidth
assigned during congestion [17].
Fig. 3.1 Architecture Diagram
For CBWFQ, packets belonging to a specific class are given
weights based on the bandwidth assigned to the class when
configured. Therefore, the order in which packets are to be
sent is determined by the bandwidth. Fair service is given to
all packets based on weight.
Strict PQ allows delay-sensitive data such as voice to be de-
queued and sent before packets in other queues are de-
queued.
Thus by the proposed queuing structure the delay can be
reduced further and the delay sensitive applications can be
supported.

_______________________________________________________________________________________
4. SIMULATION RESULTS
Fig. 4.1 Packet Delivery Ratio
Fig. 4.2 Performance Based on SNR
Fig. 4.3 Throughput Rate
5. CONCLUSIONS
In this paper, NUM problem is formulated to consider the
joint congestion control and scheduling problem for multi-
hop multipath per-link queuing wireless networks with QoS
constraints. The NUM problem is decomposed and to solve
the congestion control problem, a distributed sliding mode
controller is designed to provide multipath rate adaptation to
satisfy QoS constraints. Because of the multipath load
balancing feature, it is robust against network anomalies
such as link failure. Moreover queuing technique based on
Class Based Weighted Fair Queuing is used and priority is
added in order to schedule the packets of delay sensitive
applications.
REFERENCES
[1]. Zongrui Ding and Dapeng Wu, “Sliding Mode Based
Joint Congestion Control and Scheduling for Multiclass
Traffic Over Per-Link Queuing Wireless Networks.”
IEEE/ACM Trans. on vehicular technology, vol. 62, no. 3,
march 2013.
[2]. Yung Yi, Member, IEEE, and Sanjay Shakkottai,
Member, IEEE “Hop-by-Hop Congestion Control Over a
Wireless Multi-Hop Network”, IEEE/ACM transactions on
networking, vol. 15, no. 1, february 2007.
[3]. Zongrui Ding and Dapeng Wu, “Sliding Mode Based
Joint Congestion Control and Scheduling in Multi hop Ad
Hoc Networks with Multi-class Services”
[4]. Constantino M. Lagoa, Member, IEEE, Hao Che, and
Bernardo A. Movsichoff, Student Member, IEEE “Adaptive
Control Algorithms for Decentralized Optimal Traffic
Engineering in the Internet”, IEEE/ACM transactions on
networking, vol. 12, no. 3, june 2004.
[5]. Daniel P. Palomar, Member, IEEE, and Mung Chiang,
Member, IEEE. “A Tutorial on Decomposition Methods for
Network Utility Maximization”, IEEE journal on selected
areas in communications, vol. 24, no. 8, august 2006.
[6]. Haozhi Xiong, Ruogu Li, Atilla Eryilmaz, and Eylem
Ekici “Delay-Aware Cross-Layer Design for Network
Utility Maximization in Multi-Hop Networks”, IEEE journal
on selected areas in communications, vol. 29, no. 5, may
2011.
[7]. M. Neely, “Universal scheduling for networks with
arbitrary traffic, channels, and mobility,” in Proc. IEEE 49th
Conf. Decision Control, Atlanta, GA, 2010, pp. 1822–1829.
[8]. L. Chen, S. Low, M. Chiang, and J. Doyle, “Cross-layer
congestion control, routing and scheduling design in ad hoc
wireless networks”, in Proc. IEEE 25th Int. Conf. Comput.
Commun., Barcelona, Spain, 2006, pp. 1–13.
[9]. Nannan Zhang, Georigi M. Dimirovski, Yuanwei Jing,
and Siying Zhang, “AQM Algorithm Based on Kelly’s
Scheme Using Sliding Mode Control” American Control
Conference june 2009.
[10]. L. Georgiadis, M. Neely, M. Neely, and L. Tassiulas,
“Resource Allocation and Cross Layer Control in Wireless
Networks”. Boston, MA: Now, 2006.
[11]. D. Palomar and M. Chiang, “Alternative distributed
algorithms for network utility maximization: Framework
and applications,” IEEE Trans. Autom. Control, vol. 52, no.
12, pp. 2254–2269, Dec. 2007.

_______________________________________________________________________________________
[12]. M. Neely, “Dynamic power allocation and routing for
satellite and wireless networks with time varying channels”,
Ph.D. dissertation, Mass. Inst. Technol., Cambridge, MA,
2003.
[13]. Z. Ding and D. Wu, “Capacity region and dynamic
control of per next-hop queueing wireless networks under
time varying channels”, in Proc. INFOCOMM, 2011, pp. 1–
10.
[14]. X. Lin and N. Shroff, “Joint rate control and
scheduling in multihop wireless networks”, in Proc. IEEE
43rd Conf. Decision Control, Paradise Island, Bahamas,
2004, pp. 1484–1489.
[15]. Y. Yi and M. Chiang, “Stochastic network utility
maximization-attribute to Kelly’s paper published in this
journal a decade ago”, Eur. Trans. Telecommun., vol. 19,
no. 4, pp. 421–442, Jun.2008.
[16]. M. Neely, E. Modiano, and C. Rohrs, “Dynamic power
allocation and routing for time-varying wireless networks,”
IEEE J. Sel. Areas Commun., vol. 23, no. 1, pp. 89–103,
Jan. 2005.
[17]. Kampong J. Mabe, “LLQ in Integrated Services
Networks”, Rand Afrikaans University, Auckland Park,
South Africa.
[18]. R. Barzamini1 Student Member, IEEE, M. Shafiee2,
Member, IEEE, “A New Sliding Mode Controller for TCP
Congestion Control”, Journal. Basic. Appl. Sci. Res.,
2(7)7175-7180, 2012
[19]. Fernando Paganini, Member, IEEE, Zhikui Wang,
Student Member, IEEE, John C. Doyle, and Steven H. Low,
Senior Member, IEEE, “Congestion Control for High
Performance, Stability and Fairness in General Networks”,
IEEE/ACM Transactions on Networking, vol. 13, no. 1,
February 2005.

Congestion control based on sliding mode control and scheduling with prioritized queuing for wireless networks

Recommended

More Related Content

What's hot (17)

Viewers also liked (20)

Similar to Congestion control based on sliding mode control and scheduling with prioritized queuing for wireless networks (20)

More from eSAT Publishing House (20)

Recently uploaded (20)

Congestion control based on sliding mode control and scheduling with prioritized queuing for wireless networks