Sie sind auf Seite 1von 5

Routing and Spectrum Decision in Single

Transceiver Cognitive Radio Networks


Muhammad Zeeshan, Kashif Sattar, Zawar Shah, Imdad Ullah
School of Electrical Engineering and Computer Science, NUST, H-12 Islamabad, Pakistan
{muhammad.zeeshan; kashif.sattar; zawar.hussain; imdad.ullah}@seecs.edu.pk
Abstract Cognitive radio improves spectrum utilization
by allowing secondary users to transmit on vacant
spectrum band allocated to licensed users without affecting
their activity on that spectrum. Cognitive radio address
the problem of spectrum inefficiency as well as spectrum
scarcity but its quite challenging to capture the volatile
spectral nature of cognitive radio environment using single
transceiver. This work proposes an optimization model for
combined spectrum decision and routing in single
transceiver cognitive radio networks for maximizing
overall network capacity. Proposed centralized algorithm
ensures some minimum throughput for all flows in the
network using fairness ratio and maximizes this ratio to
increase the overall network capacity. Proposed model
considers channel heterogeneity which is a critical aspect
in making efficient spectrum decision and selects optimal
route based on primary users activity and application
requirements of the cognitive users. Results show that
proposed optimization model increases the overall network
capacity by maximizing fairness ratio for all the flows in
the network and is quite comparable with the existing
multipath routing models for cognitive radio network and
performs better in case of random topology.
Index TermsCognitive radio (CR), Channel
heterogeneity, Primary User (PU), Interference constraints
and Fairness ratio.
1.

INTRODUCTION

Cognitive radio network targets spectrum underutilization


in licensed bands by allowing secondary nodes to access
unused spectrum holes without affecting the transmission of
primary licensed users. Cognitive radio technology addresses
the two most critical problems in wireless communication
including spectrum scarcity and spectrum inefficiency.
Framework for cognitive radio allows a secondary user to
search licensed spectrum for underutilized holes and use these
spectrum according the application requirements of cognitive
users. All measures are taken to ensure that transmissions and
spectrum usage of the licensed user is not affected by
secondary users activity [1, 3].
Most devices currently being used in commercial market
have single transceiver and follow 802.11 DCF as MAC
protocol. There is need to develop comprehensive spectrum
sharing and routing protocols for these existing devices to be
used in cognitive radio environments and utilize the spectrum
more efficiently. Recent research in cognitive radio mainly
focus on spectrum sensing, finding spectrum holes, making
decision on using these spectrum holes based on primary users

activity and cognitive users suitability but all these assume the
cognitive devices to have multiple transceivers and
homogenous channels.
Channels are homogeneous in multichannel multi radio
networks and have same transmission range. Most of the work
on cognitive radios considered channels with homogeneous
behavior [2, 4 and 5, 6, 7 and 8, 9, 10]; while in practice
cognitive radio environment, channels are available on a broad
range of spectrum, and therefore channels in CR have different
properties which poses a unique challenge of finding different
neighbors each time a CR node switches channel. Depending
on their properties, channels in cognitive radio exhibit different
transmission/interference ranges, different power consumption
and different data rate supported. Heterogeneity plays a critical
role in channel assignment and its consideration makes
spectrum decision more efficient [11]. In our earlier work [12],
we evaluated solution of keeping backup channel to recover
from channel unavailability at MAC level and cooperative
channel switching to avoid problems due to channel
heterogeneity.
Due to spectrally dynamic nature of cognitive radio
networks, there is possibility of selecting same spectrum on
consecutive hops which requires time sharing and results in
decreased throughput and increased end to end delay.
Therefore routing in cognitive radio is very much dependent on
spectral information available at secondary nodes [1 and 3]. In
proposed algorithm, we assume central server or a master node
which is able to communicate with cognitive nodes around it
using a central or distributed control channel. Central node
coordinates spectrum availability information to other nodes
after making routing decision for them. Spectrum decision
mainly depends on preventing primary user activity in that area
through circumference and spectrum availability between
secondary nodes. Routing and spectrum decision in cognitive
radio becomes more challenging while considering channel
heterogeneity, previous work for single transceiver did not
consider channel heterogeneity for routing and channel
selection in cognitive radio.
In this work, we are making more realistic assumption to
create a practical cognitive environment for practical
evaluation of proposed optimization model. We are assuming
node locations, set of available channels, centralized CR server
and bi-directional links in for cognitive radio networks. These
assumptions enabled us to create a realistic cognitive
environment; link bi-directionality satisfied the 802.11 DCFs
requirement, assuming node location is more realistic in
cognitive radio networks where mobility is referred to as
switching spectrum. We propose an optimization model that
considers both routing and spectrum decision as a mixed

978-1-4673-6187-3/13/$31.00 2013 IEEE

integer linear programming (MILP) optimization problem [13].


MILP problems are NP hard so we relaxed the decision
variable to have fractional value between zero and one.
Optimization model considers channel heterogeneity while
making spectrum decision, resolves spectrum conflicts between
contending secondary nodes and prevents primary users
activity in the area by circumference. Proposed model
objectively maximizes overall network capacity using fairness
ratio between all flows in the network and ensures some
minimum throughput to each flow demand. We show that our
optimization model increases the overall network capacity by
maximizing the fairness ration for all the flows in the network.
Proposed model is quite comparable with multipath model
presented in [14] and performs better specifically in case of
random topology.
Organization of the paper is; Section 2providesoverview of
related literature. Section3 contains optimization model for
spectrum sharing and routing for cognitive nodes with single
transceiver. In Sections 4, performance of optimization model
is evaluated through analytical results. Section 5concludesthis
work and also discusses future work.
2. RELATED WORK
The concept of routing defined in cognitive radio refers to
combined selection of spectrum and route for different flows in
the network. Most of the recent work [2, 4, 5 and 6, 7, 8 and 9,
10] on spectrum decision and routing algorithms for cognitive
radio consider multiple transceivers and dedicates one
transceiver for central control channel and synchronization is
maintained to avoid deafness as proposed in [12, 15]. Currently
multiple transceiver devices are not in common use and for
making a transition from single to multiple transceiver devices
we need sophisticated routing algorithms to make this
technology shift productive, smooth and economical. Channel
heterogeneity, a must consideration in cognitive radio
environment is also ignored in most recent work [2, 4, 5 and 6,
7, 8, and 9, 10] and is only considered in [11].Exploitation of
channel heterogeneity is must in cognitive radio considering
the fact that spectrum availability in cognitive radio is sporadic
and different spectrum band have different range and support
different data rate as handled in [11].Existing work is classified
in [3] on the basis of functional support provided by proposed
algorithms. There are algorithms that provide channel selection
as well as the selection of next hope forwarding node [5, 6],
some jointly make spectrum decision with awareness of
primary users activity [4,9] and others provide reconfigurability and route recovery along with routing and
spectrum decisions [8, 10].
Work in [16] presents energy aware routing and spectrum
selection for single transceiver cognitive radios, they proposed
algorithm for routing and time-slot assignment with an
objective of balanced energy consumption, contention
elimination between cognitive users and decomposition of
different traffics flows to different frequency time-slots. In
[11], authors critically modeled spectrum selection,
interference and heterogeneity by formulating an optimization
model. Results showed that channel heterogeneity plays a
critical role in optimal channel assignment solutions and we
have also catered channel heterogeneity in our cross-layer
optimization model.
In [14] authors presented a cross-layer optimization
framework considering routing and spectrum sharing for
maximizing fairness ratio. But their optimization model is
generic to multi transceiver with multipath routing in cognitive

radio networks while our work is more realistic and focuses on


single-transceiver with unipath routing. Our work on being
single transceiver has its unique importance of being widely
deployed in todays wireless networks that run 802.11 DCF
MAC with the assumption that cognitive node knows the
control channel and switches/listens to control channel after
every specified period for maintaining synchronization
between other nodes [12, 15]. We considered 802.11-style
interference model as evaluated in [17]. RTS/CTS method was
also recommended by Giuseppe Bianchi for majority of
practical wireless network deployments because RTS/CTS
method solves the hidden node problem and its advantages in
large network scenarios [18].Routing algorithms for single
transceiver cognitive nodes proposed in MSCRP[7],channel is
selected based on switching delay, contention for that channel
and data transmission. But flooding of RREQ to all available
channels is not a scalable solution for larger networks. G.
Cheng et al. used the modified AODV for route discovery,
developed multi-flow multi-frequency spectrum scheduling
scheme and analyzed all possible delays during spectrum
decision for cognitive radios [19]. Path in [7, 19] is chosen
before evaluation of spectrum availability on that path for
generic metrics including end to end delay and hop count,
which certainly is not a practical solution.
3.

CROSS-LAYER DESIGN OF ROUTING AND SPECTRUM


DECISION

We formulated cross layer optimization framework for


combined decision on routing and spectrum decision
considering the heterogeneous channels on each node.
Spectrum sharing refers to the decision of assigning specific
frequency bands to active links. Our aim in this model is to
maximize network capacity by exploiting channel diversity and
allocating bandwidth fairly through a fairness ratio. We also
ensure a conflict free topology through interference and
unipath routing constraints. Notations are listed in Table 1.
Table 1 Notations
Symbol
Description

Fairness ratio
Set of channels
All nodes in the network
Neighboring nodes of node
All links available in the network
,
,
,

Available channels at node


Link between nodes and node
Common channels available between node and
Interfering links with link , on channel
All flows available in the network
Set of Intermediate nodes for flow
Assignment (0,1) of channel on link , for flow
Source node for flow
Destination node for flow
Flow demand for flow
Neighboring nodes of
Neighboring nodes of

Bandwidth of channel

over link

3.1. Assignment on links


For reliable transmission, two cognitive nodes must be
within single transmission range and must be tuned to same
frequency band i.e., common channel. A link , is only

active if two nodes are tuned to same channel


over
link
, . We define a binary decision variable
, where
is one of the common channels available
between node and i.e.,
,
:
,
0,1
(1)
, , ,
3.2. Interference constraint
Interference model of 802.11 defines that if a link i, j is
assigned and active on channel m, then channel m must not be
i, j
assigned to any other link p, q as long as p, q
and
i, j is the set of interfering links with the link i, j on
channel m . As per our assumption of bi-directional links,
traffic on each link is flowing in both directions. Therefore one
thing is important here that link j, i can also be included in
the set of
i, j .
,
,
,
1
(2)
,
,
,
,
3.3. Non-negativity
This constraint prevents the flows to have less than zero
demand on all links. Due to our assumption of bi-directional
links, our model ensures non-negativity in both directions:
,
,

0,
0
(3)
,
,
,
,
3.4. Unipath routing
We define flow demand for each source destination pair
and use f (f = 1, 2, F ) for flow indexing. Notations
,
represent source node, destination node and
demand for a particular flow . Where i, j
E i.e., set of all
link available in network and i j. We have following unipath routing constraints:
3.4.1. Source Flow
Source flow constraint makes sure that each flow generated
from any source in the network will go through only one link
towards the destination:

,
1
(4)
3.4.2. Flow conservation
This constraint ensures that on each node the amount of
incoming flow is exactly equal to outgoing flow, except source
and destination nodes i.e., on intermediate nodes only. This
constraint makes sure that there is no loss on the intermediate
nodes:

,
,
(5)
,
3.4.3. Destination flow
Destination flow constraint ensures that the flow generated
equals to flow destined:

,
,
(6)
3.4.4. Non reversive flow
This constraint ensures that the flow will not come back on
the same link once forwarded:

,
,
1
(7)
,
,
3.5. Link capacity
This constraint restricts that sum of all flows passing
through a link , on a particular channel m in both directions
must not exceed the link capacity:

,
,

,
(8)

,
,
,
3.6. Problem formulation
For imposing fairness criteria on each flow demand,
objective function implements the scaling factor for all flows
in network and this satisfies our goal to maximize the overall
network capacity and ensuring fairness among all flows in the
network. Following is the problem formulation:

max
,

,
0, 1
subject to
constraints (2) to (8)
, is abinary integer. To make
The decision variable
sure that the problem is solved in polynomial time, we allowed
, to have value either 0 or 1. The
the decision variable
value of varies between 0 and 1. After solving the relaxed
problem, we are able to obtain upper bound of optimum results.
This implements fairness over all the flows in network and
maximizing fairness ratio will maximize the effective
throughput of the network.

4. RESULTS AND DISCUSSION


We evaluated two kind of topologies with respect to node
placement i.e. grid topology and random topology and in each
topology we considered four flows.
4.1. Performance (grid topology)
Grid topology consists of 12 nodes placed in 3x4 grids;
total area in which nodes are deployed is in square size of 80m
x 80m and the distance between each node is 20m. There are
12 channels available in the network distributed into three
different ranges i.e., 10m, 20m and 30m (Table 2) and Table 3
shows channels availability on nodes.
Table 2 Heterogeneous transmission
Transmission range
Available channels
10
10, 7, 2
20
12, 11, 8, 4, 3
30
9, 6, 5, 1

Figure 1. Connectivity graph for grid topology

Connectivity graph in Fig.1 shows connections between


nodes for grid topology and there are 19 available links.
Existence of the link between two nodes shows that these
nodes have at least one common channel for this range. For
example there is a common channel between nodes 1 and 2 and
also a common channel between nodes 3 and 4. Table 4 shows
the flow details for one of experiment including sourcedestination pair and demand rate of flow between them, these
values are generated randomly for each experiment.
Table 3 Set of available channel (i.e.,
) for grid topology
Node
Coordinates (location)
Channels
1
20,20
3,4,6,8

2
3
4
5
6
7
8
9
10
11
12

20,40
20,60
20,80
40,20
40,40
40,60
40,80
60,20
60,40
60,60
60,80

4,7,11
1,2,5,6,12
4,5,8,10
3,5,8,10,11
2,3,5,7,9
1,2,6,7,9,12
2,4,6,11,12
2,4,5,6,7,11
3,6,9,11
1,5,7,8
2,4,8,10

Table 4 Source, destination and rate for flows in grid topology


Source
Destination
Rate demand
11
8
9
1

12
10
3
7

135
142
156
140

Table 6 Set of available channel for Random Topologies


Node
Location
Location
Available
Topology1
Topology2
Channels
1
16,31
20,10
3,4,6,8
2
7,22
42,45
4,7,11
3
33,26
28,5
1,2,5,6,12
4
35,45
29,31
4,5,8,10
5

45,49

10,25

3,5,8,10,11

8,42

6,34

2,3,5,7,9

20,21

9,11

1,2,6,7,9,12

39,17

42,29

2,4,6,11,12

41,32

32,20

2,4,5,6,7,11

10

35,0

34,47

3,6,9,11

11

1,8

21,24

1,5,7,8

12

23,8

15,42

2,4,8,10

Fig.2 shows comparison of proposed unipath optimization


model with the results obtained from implementing multipath
optimization model presented in [14] for grid topology. The
dashed line for each experiment show higher values as model
in [14] is multipath and utilizes multipath using multiple
transceivers for different flows while our model have limitation
of single path using single transceiver per cognitive node and
still shows quite comparable results with the multipath model.
Table 5 shows the gap between two values and this can be seen
that gap in nominal taking in consideration the difference of
single and multipath routing models and single and multiple
transceivers at cognitive nodes.

Figure 3. Connectivity graph for Random Topology-1

Figure 4. Comparison of fairness ration () between uni-path and multipath


[14] models for random Topology-1

Figure 2. Comparison of fairness ration () between uni-path and multipath


[14] models for grid topology

Table 5 Performance gap for grid topology


Gap ( Multi Unipath)
Experiment
1
(0.93 - 0.46) = 0.47
2
(0.86 - 0.43) = 0.43
3
(0.91 - 0.45) = 0.46
4
(0.69 - 0.34) = 0.35
5

(0.76 - 0.38) = 0.38

4.2. Performance (Random topology)


Now we evaluate performance of our model with two
different random topologies one of them is shown in Fig.3, we
took 12 nodes placed at randomly generated locations given in
Table 6 and nodes are deployed in 50m x 50m area. For
Topology-1, 12 channels available in network distributed into
two different ranges i.e., 20m and 30m. Channels 1-6 have 20m
transmission range and channels 7-12 have transmission range
of 30m whereas for Topology-2 we considered the same ranges
and channels as in Table 2.

Fig.4 and Fig.5 show comparison of our proposed singlepath optimization model with the results obtained from
implementing multipath optimization model presented in [14]
for random Topology-1 and Topology-2 respectively.
Similarly here the dashed line for each experiment show higher
values, as model in [14] is multipath and utilizes mutipath
using multiple transceivers for different channels while our
model have limitation of single path using single transceiver
per cognitive node and still shows quite comparable results
with the multipath model. Table 8 shows the gap between two
values of each random topology. It can be seen that gap is
nominal, taking in consideration the difference of single and
multipath routing models and single and multiple transceiver at
cognitive nodes.
Table 7 Source, destination and rate for random topology-1
Source
Destination
Rate demand
11
6
4
9

5
10
2
12

168
183
175
148

Table 7 shows the flow detail for one of experiment


including source-destination pair and demand rate of flow
between them.

Results of our proposed optimization model for spectrum


sharing and routing show increase in the overall network
capacity by maximizing fairness ratio for all the flows in the
network and our model caters for channel heterogeneity as a
generic property of cognitive radio. We compare the results of
our unipath routing with the multipath routing model and show
that proposed model performs comparably to multipath routing
model in [14] and achieves similar fairness ratio especially for
random topologies. In our future work we are working on
modeling distributed algorithm for spectrum selection and
routing in cognitive radio networks with single transceiver.
REFERENCES

Figure 5. Comparison of fairness ration () between uni-path and multipath


[14] models for random Topology-2

Comparing both results for grid and random topologies, In


grid topology optimization for multipath routing perform much
better than proposed model but keeping in mind the differences
between both models we have achieved good fairness ratio
with increased overall network throughput for all the flows in
network. While in random topologies, our optimization model
shows very good results as at times it even matches the fairness
ratio achieved by multipath routing model and indeed its quite
interesting to see such results in cognitive environment in
which all cognitive nodes are equipped with only one
transceiver. Fig.6 shows the comparison of all the three
topologies which we have considered for our results. The
results of our unipath optimization model are very close to each
other for each kind of topology which also proves the
correctness of our model.
Table 8 Performance gap for random topology
Experiment
Topology-1
Topology-2
Gap ( Multi Unipath)

Gap ( Multi Unipath)

(0.46 - 0.33) = 0.13

(0.38 - 0.38) = 0.00

2
3
4

(0.29 - 0.29) = 0.00


(0.38 - 0.25) = 0.13
(0.31 - 0.31) = 0.00

(0.36 - 0.32) = 0.04


(0.40 - 0.35) = 0.05
(0.43 - 0.34) = 0.09

(0.38 - 0.33) = 0.05

(0.37 - 0.32) = 0.05

Figure 6. Comparison of different topologies with our proposed Uni-path


optimization model

5. CONCLUSION
Previous work on single transceiver in cognitive radio
network ignored the aspect and consequences due to channel
heterogeneity and only considered homogenous channels. We
proposed an optimization algorithm for combined routing and
channel assignment in cognitive radio network with single
transceiver for maximizing overall network throughput. Our
proposed model caters the effects due channel heterogeneity
for making efficient spectrum decision and selects optimum
route while preventing primary users activity and also ensures
fairness among flows of secondary nodes in the network.

[1] Marinho, J., & Monteiro, E, Cognitive radio: Survey on communication


protocols, spectrum decision issues, and future research directions, Wireless
Networks, 18(2), 147164, 2012.
[2] Angela Sara Cacciapuoti, Marcello Caleffi, Luigi Paura, Reactive
routing for mobile cognitive radio ad hoc networks, Ad Hoc Networks,
Special Issue on Cognitive Radio Ad Hoc networks, Volume 10, issue 5, pp
803 815, 2012.
[3] I. F. Akyildiz, W. Y. Lee, and K. R. Chowdhury, CRAHNs: Cognitive
Radio Ad Hoc Networks, in Elsevier Ad Hoc Networks Journal, vol. 7, no. 5,
pp. 810-836, July 2009.
[4] C.W. Pyo, M. Hasegawa, Minimum weight routing based on a common
link control radio for cognitive wireless ad hoc networks, in: Proceedings of
the International Conference on Wireless Communications and Mobile
Computing (IWCMC), August 2007, pp. 399404.
[5] H. Khalife, S.S. Ahuja, N. Malouch, M. Krunz, Probabilistic path
selection in opportunistic cognitive radio networks, in: Proceedings of the
IEEE Globecom, November 2008.
[6] H.-P. Shiang and M. van der Schaar, Delay-sensitive resource
management in multi-hop cognitive radio networks, in Proc. IEEE DySPAN,
October 2008.
[7] H. Ma, L. Zheng, X. Ma, Y. Luo, Spectrum-aware routing for multi-hop
cognitive radio networks with a single transceiver. In Proc. of the Cognitive
Radio Oriented Wireless Networks and Communications (CrownCom), May
2008.
[8] G.-M. Zhu, I. Akyildiz, and G.-S. Kuo, STOD-RP: A Spectrum-Tree
Based On-Demand Routing Protocol for Multi-Hop Cognitive Radio
Networks, in IEEE GLOBECOM 2008, Dec. 2008, pp. 15.
[9] Chunsheng Xin Liangping Ma Chien-Chung Shen, A Path-Centric
Channel Assignment Framework for Cognitive Radio, Wireless Networks,
Springer July 2008.
[10] A. Sampath, L. Yang, L. Cao, H. Zheng, and B. Y. Zhao, High
Throughput Spectrum-aware Routing for Cognitive Radio based Ad-hoc
Networks, in CrwownCom 2008.
[11] M. Ma and D. Tsang, Impact of channel heterogeneity on spectrum
sharing in cognitive radio networks, in Proc. IEEE ICC, 2008, pp. 2377
2382.
[12] Muhammad Zeeshan, Muhammad Fahad Manzoor, Junaid Qadir,
Backup Channel and Cooperative Channel Switching On-Demand Routing
Protocol for Multi-Hop Cognitive Radio Ad Hoc Networks (BCCCS), 6th
IEEE International Conference on Emerging Technologies 2010 (ICET 2010).
[13] R. Fourer, D. M. Gay, and B. W. Kernighan, AMPL, A modeling
language for mathematical programming, 1993, http://www.ampl.com.
[14] Miao Ma, Danny H.K. Tsang, Joint design of spectrum sharing and
routing with channel heterogeneity in cognitive radio networks, in Physical
Communication 2, 2009.
[15] J. So, N. Vaidya, Multi-channel MAC for ad hoc networks: handling
multi-channel hidden terminals using a single transceiver, in: ACM
International Symposium on Mobile Ad Hoc Networking and Computing
(MOBIHOC), May 2004, pp. 222233.
[16] S. M. Kamruzzaman, Eunhee Kim, Dong Geun Jeong, Spectrum and
Energy Aware Routing Protocol for Cognitive Radio Ad Hoc Networks,
IEEE International Conference on Communications ICC 2011.
[17] P. Gupta, P.R. Kumar, The capacity of wireless networks, IEEE
Transactions on Information Theory 46 (2) (2000).
[18] Giuseppe Bianchi, Performance Analysis of the IEEE 802.11
DistributedCoordination Function, IEEE Journal on Selected Areas in
Communications, VOL. 18, No. 3, March 2000.
[19] Cheng, W. Liu, Y. Li, and W. Cheng, Spectrum Aware On-Demand
Routing in Cognitive Radio Networks, in IEEE DySPAN 2007, April 2007,
pp. 571574.

Das könnte Ihnen auch gefallen