Cross Layer Routing for Multihop Cellular
Networks
G. Kannan, S. N. Merchant and U. B. Desai
SPANN Laboratory, Electrical Engineering Department, IIT Bombay, India
Email: {gkannan, merchant, ubdesai}@ee.iitb.ac.in
Abstract— We propose a unified cross layer routing protocol with multiple constraints for CDMA multihop cellular
networks (MCN). Multiple constraints are imposed on intermediate relay node selection and source to destination path
selection. The relay node constraints for routing protocol design are cooperation, interference caused to other nodes and
suf f icient neighborhood connectivity. Path constraints for
routing are end-to-end throughput and end-to-end delay. We
do not assume full cooperation for call forwarding and present
a facile incentive mechanism to motivate the cooperation for
call forwarding. We incorporate a realistic mobility model and
dynamic call dropping notion in our system. Flat fading i.i.d
Rayleigh channel is assumed between mobile nodes. Simulation
results show that the proposed algorithm reduces intracell interference by more than 70% as compared to nearest neighbour
routing algorithm. Furthermore, the proposed algorithm achieves
better throughput and maintains the delay at tolerable level. We
also demonstrate that the proposed routing algorithm is superior
to many standard routing protocols.
2)
3)
4)
I. I NTRODUCTION
In conventional cellular networks (2G, 2.5G, 3G), all communication go through a base station in single hop fashion.
However, recent studies [1], [2], [3] have indicated that there is
indeed an improvement in performance, especially in terms of
the capacity, coverage enhancement and power utilization, by
using MCN. Though MCN leads to performance enhancement,
there are a number of issues that have to be addressed to make
MCN an effective technology [1]. We address one such important issue known as routing. Most of the protocols currently
are designed to optimize one or more of the following metrics:
energy, delay, distance, end-to-end throughput and interference
[4], [5], [6], [7]. Nevertheless, we believe that there is a need to
design a global optimal routing protocol for MCNs by taking
all necessary issues into considerations to maximize the utility.
We propose a cross layer routing protocol with multiple
constraints for CDMA multihop cellular networks. The multiple constraints are divided into node constraints and path
constraints. The node constraints are cooperation, interference
level and suf f icient neighborhood connectivity. The constraints for path selection are end-to-end throughput and endto-end delay. Our work considerably differs from most of the
existing work on routing protocols in the following ways:
1) We introduce suf f icient neighborhood connectivity
as a new routing metric. In other words, the nodes which
This work was supported by Microsoft Corporation and Microsoft Research
India under the Microsoft Research India PhD Fellowship Award.
5)
have sufficient surrounding neighbors are selected as
relay nodes rather than isolated nodes. This new metric
will be highly beneficial when the call is dropped due
to various reasons in between communication.
We incorporate the notion of dynamic call dropping in
our model. We allow the participating nodes in routing to
drop the call in between communication. Call dropping
may be due to the mobility of relay nodes or the urgency
of a relay node to make its own call, etc. Hence, when
the call is dropped at any particular node, one of the
neighbor around that particular node will be selected as
relay node.
We do not assume full cooperation from nodes for call
forwarding and propose a novel incentive mechanism to
encourage the cooperation.
We formulate the end-to-end throughput by assuming
i.i.d Rayleigh flat fading propagation channel between
nodes instead of the commonly followed distance-decay
law in literature.
We combine multiple metrics in a meaningful way to
find an optimal cross layer routing protocol in MCNs.
We design the routing protocols by taking into account
the propagation channel conditions and MAC protocol
interference levels.
II. R ELATED WORK
MCN architecture introduced in [1] uses shortest path
algorithm for routing. Routing protocol for hybrid networks
based on spanning tree algorithm is proposed in [2]. A relay
node selection based on a route that has the lowest bottleneck
in a two-hop relaying network is proposed in [3]. However,
both [2] and [3] do not guarantee a sufficient quality of service
(QoS) and assume fully cooperating nodes. A charging and
rewarding policy in routing for MCN is proposed in [8] but the
routing protocol used is DSR. A routing selection based on call
status, signal strength, battery power and round trip time has
been proposed in [9]. Many power-aware routing protocols that
consider energy as the metric have been proposed in literature
for ad hoc networks [4], [10], [11], [12]. All the above
references assume distance-decay law as a network model
which is highly impractical in MCN because fading factor
is prevalent. Routing for ad-hoc networks with transmission
energy as criterion for Rayleigh fading channel is proposed
in [13], but this model assumes zero interference and full
cooperation. To the best of authors’ knowledge, there is no
21st International Conference on
Advanced Information Networking and Applications Workshops (AINAW'07)
0-7695-2847-3/07 $20.00 © 2007
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on December 3, 2008 at 00:29 from IEEE Xplore. Restrictions apply.
IV. N ODE METRICS FORMULATION
The following are the metrics in selecting the relay nodes
from the set of n nodes
1) Cooperation of the nodes.
2) The interference caused by the call forwarding nodes.
3) Connectivity of the nodes.
Mobile terminals
A. Cooperation metric and Proposed Incentive Mechanism
Base station
Single hop
control messages
Multihop Voice/
Data messages
Fig. 1.
Proposed routing architecture
literature available on optimum routing with multiple QoS
constrains which maximizes the throughput in Rayleigh fading
channel model without assuming full cooperation. Moreover,
there is no analysis done on dynamic call dropping.
III. S YSTEM MODEL AND A SSUMPTIONS
We consider a single cell with single base station at the
center and n nodes distributed according to two dimensional
Poisson point process. We assume that the base station and
nodes use CDMA as access technique for their interconnections [6]. Perfect power control is assumed for this CDMA
network, so that all the transmitters use just the transmission
power level that is required to let the receiver decode the
signal with proper quality. However the nodes are assumed
to transmit a minimal power pmin when its intended receiver
is at a very less distance. The communication is assumed
to be in the form of packets. We assume that nodes do not
transmit and receive at the same time slot to avoid collision
at nodes. The relay nodes are assumed to be of despread and
forward type: i.e relay nodes despread the CDMA packet and
spread them again with the different PN code and forward
it to the next node in the path. We assume that each node
independently decides to participate in call forwarding or
otherwise. The logical channel is divided into control channel
(CCH) and traffic channel (TCH). Note that the logical channel
is different from propagation channel. CCH handles only
signaling, while TCH carries speech and data traffic. Control
messages containing the source ID and the destination ID, will
be exchanged between the nodes and base station using CCH.
Control messages follow single hop communication as they
are very short and occur, only at the time of call initialization.
TCH follows multihop communication. The route for TCH
from source to destination will be found out by base station
using the proposed routing protocol. This route map will be
conveyed to source using single hop communication through
CCH.
We select nodes which can cooperate for call forwarding as relay nodes. We propose an incentive mechanism
to stimulate the node cooperations as follows: whenever a
node wants to initiate a communication, it will send a call
initiation request to the base station through CCH. Upon
receiving the call initiation request, base station will broadcast a cooperation request to the whole network through
broadcast CCH. The cooperation request contains source
ID, destination ID and the incentive amount per node that
will be paid after communication. Now those nodes which are
interested in incentives reply back to the base station using
CCH. Let us assume Φ(n̄) is the network of cooperating nodes.
There are many solutions available in the mode of paying
incentives to the user [14], [15]. The incentives amount can
also be chosen adaptively until Φ(n̄) reaches sufficient density.
The incentives for multihop routing will be paid by service
provider as the service provider is beneficiary in MCN.
B. Interference metric
Interference reduction will be achieved by controlling the
transmitted power. For a particular node i, we shall find the
next intermediate node j such that the communication between
them causes minimal interference in the network.
Let us consider the communication between a particular
node i and any other node j as shown in Fig. 2. The average
interference received at some node
r due to transmission from
ρ2 p
node i to node j is given by irdγ ij where ρir is the timeir
correlation between the signature waveforms of nodes i and
r, γ is the path loss coefficient, dir is the distance between
node i and node r and pij is the transmitted power from
node i to node j. Note that average received power in fading
channel always follows distance-decay law [16]. Let us assume
G is the DS-CDMA modulation processing gain, sum of the
interference received in all neighbor nodes in the network due
to the transmission from i to j is given by
I=
1
G
n
r=1,r={i,j}
ρ2ir
pij dγir
Assume that the transmitted power levels at and node i is
adjusted such that the destination node j receives a power
of pref i.e pij = pref dγij Now the interference received at
neighbor nodes as a function of dij is
1
I(dij ) =
G
n
p dγ
2 ref ij
ρir
dγir
r=1,r={i,j}
21st International Conference on
Advanced Information Networking and Applications Workshops (AINAW'07)
0-7695-2847-3/07 $20.00 © 2007
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on December 3, 2008 at 00:29 from IEEE Xplore. Restrictions apply.
number of connected neighbors (connectivity criterion), whenever one or more of the above stated situation(s) arise, the
corresponding intermediate node (depleted node) informs the
base station through the CCH. Base station will pickup one
of the neighbors of the depleted node as a substitute such that
the constraints on all the metrics are satisfied. The depleted
node will be relieved subsequently. Such a route finding occurs
without breaking the ongoing call.
r
q
Interferences
djr
Interferences
diq
Intended signal
(Forwarded call) dij
i
Interferences
dis
Fig. 2.
nodes
j
s
Interference at other nodes due to communication between any two
where pref and dir are fixed quantities and the only variable
is dij . The interference caused at other nodes due to communication from j to i can also be derived in the similar
way. Let us construct a subset Φ(n̂) from Φ(n̄) such that
the communication between any two nodes in Φ(n̂) causes
minimal interference in the network.
C. Connectivity metric and Dynamic call dropping
1) Connectivity: We define connected neighbor as follows.
For a particular node i, any node j (i, j ∈ Φ(n̂)) which lies
γ1
is said to be a
at a distance dij such that dij < ppmin
ref
connected neighbor of i. In [17] it is shown that in a network
of n randomly placed nodes, each node should be connected
to Θ(log(n)) neighbors. If a node has less than 0.074(log(n))
connected neighbors, then the network is asymptotically disconnected with probability one as n increases. Furthermore,
if each node is connected to more than 5.1774(log(n)) neighbors, then the network is asymptotically connected with probability approaching one as n increases. Hence, connectivity is
an important criterion to establish communication. We define
suf f icient neighborhood connectivity of nodes as follows:
Suppose node m ∈ Φ(n̂) has k number of neighbors and if
k > Θ(log(n))
(1)
then node m satisfies suf f icient neighborhood connectivity
criterion and consequently, it is eligible for being a relay
node. Let us construct a sub set Φ(n̈) from Φ(n̂) with all such
ms.We call the nodes in Φ(n̈) as potential relay nodes.
2) Dynamic Call Dropping: The forced termination of the
call against the will of the subscriber is called dynamic call
dropping. Dynamic call dropping may be due to various
reasons. The mobility of the intermediate nodes is the major reason for call dropping in multihop cellular networks.
Moreover, there can arise an emergency situation that, the
intermediate node may break the ongoing communication and
try to make its own emergency call.
The proposed solution for dynamic call dropping is as
follows: Since every selected intermediate node has sufficient
V. PATH METRICS FORMULATION
Let X = χ1 , χ2 , χ3 , . . . , χM denote the set of paths
available between source node and destination node along
the potential relay nodes. The following are the metrics in
choosing a particular path χm from the set X
1) End-to-end throughput
2) End-to-end delay
A. End-to-end throughput metric
End-to-end throughput is defined as the probability of
successful transmission from source node to destination node
which involves successful transmission at each and every
intermediate node. The successful single hop transmission
from node i to its neighbor node j (∀i, j ∈ Φ(n̈)) occurs
when the received power at node j from node i is stronger
than interference plus noise power by a factor of β (i.e
SIN R ≥ β). The probability of successful transmission
from node i to node j is
P(Cij ) = P(SIN Rij ≥ β) = P(rij ≥ β.(Iij + N ))
where rij is the received power at node j from the intended
node i and Iij is the interference at node j due to other
communications. Let rkj , k = 1, . . . , K (k = i, j) be the
received power at node j from kth interferer. The interference
at node j from all interferers is given by
K
1
ρ2kj rkj
Iij =
G
k=1,k={i,j}
Erroneous detection occurs when SIN Rij < β, this probability P(Eij ) is given by P(Eij ) = P(rij < β.(Iij + N ))
The propagation channel between mobile to mobile is different
from the conventional wireless channel. However the envelope
still follows Rayleigh distribution [18]. By the fact that if Y is
Rayleigh distributed and X=Y2 , then X will follow exponential
−rij
distribution as follows: P(rij ) = R1ij e Rij where Rij denotes
p
the average received power Rij = dij
γ [13], [16] .
ij
Let us say χm = {1, 2, 3, . . . , h} is the path selected to
relay the communication from source node 1 to the destination
node h and number of hops in the communication is h −
1. The probability P(C1h ) that the message is successfully
transmitted from source 1 to destination h is given by
h−1
P(C1h ) = P(
h−1
Cii+1 ) =1 − P(
i=1
≥1−
Eii+1 )
i=1
h−1
21st International Conference on
Advanced Information Networking and Applications Workshops (AINAW'07)
0-7695-2847-3/07 $20.00 © 2007
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on December 3, 2008 at 00:29 from IEEE Xplore. Restrictions apply.
i=1
P(Eii+1 )
where I12 itself is a random quantity, therefore, by analyzing
along the similar lines of [16], P(E12 ) can be written as
∞
∞
2
β[ G1 K
k=3 ρk2 rk2 +N ]
R12
P(E12 ) =
1 − e−
...
×
0
K
0
1000
88
59
79
92
39
97
700
83
600
500
8
24
0
0
72
15
44
31
99
87
98
67
55
94
32
43
70
18
34
64
54
58
200
22
1
82
45
38 28
40
96
69
300
60
66
14 16
71
63
Source node
73
100
29
52
Base station
25
84
7
78
36 50
56
10 5
65
49
62
74
75
11
47
P (rk2 )drk2
26
85
30
35
91
2
51
42
61
89
41
9
46
77
48
90
200
76
4
33
300
6
27
93
3
400
100
19
23
Destination node
100
17
80
95
81
86 68
800
57
37
12
900
y coordinates of nodes in meter
where the last inequality is obtained by using the property
of union bound. Note that here P(Cii+1 ), i ∈ Φ(n̈) is
dependent on correct detection of all its previous nodes, hence
independence assumption does not hold here. Let us consider
communication between node 1 and 2.
β.(I12 +N )
r12
−β(I12 +N )
1
(e− R12 )dr12 = 1 − (e R12 )
P(E12 ) =
R12 0
400
500
600
700
x coordinates of nodes in meter
20
800
13 5321
900
1000
k=3
By substituting P(rij ) and by invoking independence of P(rij )
the above equation can be written as
K
− βN−γ
1
p12 d12
.
P(E12 ) = 1 − e
β ρ2k2 pk2 d12 γ
k=3 1 + G p12 ( dk2 )
Hence the end-to-end throughput can be written as
− βN
h−1
−γ
1 − e pii+1 dii+1 .
P(C1h ) ≥ 1 −
i=1
K
.
k=1,k={i,i+1}
1
1+
2
β ρki+1 pki+1 dii+1 γ
( dki+1 )
G
pii+1
(2)
B. End-to-end delay
The major contributions for the end-to-end delay result from
the transmission delay induced by the relay nodes and the
propagation delay over the multihop communications. In our
routing algorithm, we ensure that end-to-end delay is minimum
by involving additional path constraint.
VI. ROUTING P ROTOCOL FOR DATA / VOICE MESSAGES
1) From n nodes, select a set of cooperative nodes and
construct a set Φ(n̄).
2) From Φ(n̄) form a set Φ(n̂) consisting of nodes which
satisfy interference criterion.
3) From Φ(n̂) choose a set of nodes which have
suf f icient neighborhood connectivity and build a set
Φ(n̈).
4) From Φ(n̈) select source to destination paths X such that
the lower bound on P (C1h ) is above a certain threshold.
5) From X choose a source to destination path which has
minimal end-to-end delay.
VII. S IMULATIONS AND R ESULTS
We simulate a single-cell DS-CDMA system. The simulation parameters are presented in table below:
Fig. 3.
Various routing algorithms: Green- proposed algorithm, RedInterference aware routing, Gold- Optimum hop size routing, Black- Nearest
neighbor algorithm, Pink- Conventional communication
Parameters
Cell radius
Propagation loss exponent
SINR threshold (β)
Cooperation level
End-to-end delay threshold
Correlation coefficient
between spreading codes
Spreading factor
Thermal noise at receiver
pref
Antenna Gain in MT
Value
1 Km
4
3 dB
70 %
100 msec
0.1
32
-90 dB
10−7 watt
0 dB (Omni directional)
Interference threshold (Imax ) and end-to-end throughput
threshold, have been chosen empirically as function of number
of nodes. To validate the performance of the proposed model,
Monte-Carlo simulation was carried out by randomly selecting
the source, destination nodes and the results were averaged
over at least 500 realizations of the nodes distribution. We
compared our proposed algorithm with interference aware
routing proposed in [5], optimum hop size routing proposed
in [6] and nearest neighbor routing in underling simulation
environment. Fig. 3 gives the route map of various routing
protocols.
A. Total power analysis
The transmission power model used for analysis is pij =
αd4ij + pmin , where α is the normalization constant. For
simulation purpose, let us assume pmin = 0.1 and the
maximum transmitted power from mobile handset is 2 watts.
By substituting these values, we can rewrite the above equation
as pij = 1.9 × 10−12 × d4ij + 0.1 . Apart from the transmission
power, we also consider receive power in our power analysis.
We assume a constant receive power of 50 mwatts per
packet per node in our simulations. Fig. 4 compares the total
power spent per packet transmission in various algorithms.
21st International Conference on
Advanced Information Networking and Applications Workshops (AINAW'07)
0-7695-2847-3/07 $20.00 © 2007
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on December 3, 2008 at 00:29 from IEEE Xplore. Restrictions apply.
Nearest neighbor algorithm
Interference aware algorithm
Optimum hop size algorithm
Proposed algorithm
1.8
1.6
1.4
1.2
1
0.8
Fig. 4.
0.7
0.6
0.5
0.4
0.3
0.6
0.2
0.4
0.1
0.2
500
1000
1500
2000
Number of nodes
2500
3000
Total power spent per packet transmission versus number of nodes
25
Nearest neighbor algorithm
Interference aware algorithm
Optimum hop size algorithm
Proposed algorithm
0.8
Dynamic call droping probability
Total power spent per packet transmission in watt
2
500
Fig. 6.
1000
1500
2000
Number of nodes
2500
3000
Call dropping probability versus number of nodes
C. Call dropping analysis
Nearest neighbor algorithm
Number of nodes bathed with interference
Interference aware algorithm
Optimum hop size algorithm
20
Proposed algorithm
15
10
5
0
500
1000
1500
2000
Number of nodes
2500
3000
Fig. 5. Number of nodes affected with interference due to communication
between nodes against number of nodes
To visualize the effect of call dropping in multihop communication, we have implemented a model as given below:
1) A simple mobility model wherein at each time instant,
a user is randomly located within a circle of diameter
50 m with the center of this circle being the user’s
previous location.
2) A model in which there is a need for an intermediate
node to initiate its own communication with probability
less than 0.01.
We have chosen Θ in (1) as a linear function of nodes with
slope of 0.0025. The results are shown in Fig. 6. We can infer
that the dynamic call dropping probability is very less in the
case of proposed algorithm because of connectivity criterion.
D. Incentives and end-to-end delay analysis
From Fig. 4 we can infer that the proposed routing algorithm
decreases the required overall power to a larger extent as
compared to other algorithms. An interesting observation is
that, the total power spent on the nearest neighbor algorithm
and interference aware algorithm are considerably higher.
This is because, though the single hop transmission power
in both the algorithms is less, these algorithms result in many
intermediate nodes; hence the total transmitted power spent
will be significantly higher.
B. Interference analysis
Due to the reasons stated above the number of nodes affected with significant interference because of communication
from a source to destination is considerably lesser in the proposed algorithm as shown in Fig. 5. The nodes which receive
interference more than 0 dB from any other communication
is assumed to have significant interference. From Fig. 5 we
can also infer that the reduction in interference in the case of
proposed algorithm compared to nearest neighbor algorithm is
about 70 %.
In the end-to-end delay, transmission delay in the intermediate nodes constitutes major component. Therefore, as the
number of intermediate nodes increases the end-to-end delay
and the incentives paid also increase. We assume equal distribution of incentives for all intermediate nodes and constant
transmission delay of 30 msec per node. Fig. 7 compares
end-to-end delay and Fig. 8 compares the incentives paid in
various algorithms. From Fig. 7 we can infer that the endto-end delay in case of interference aware routing is at the
intolerable level for voice communication, while the end-toend delay in the proposed approach is around 100 msec which
is insensitive to human ears. From Fig. 8 we can conclude that
the proposed algorithm costs much less compared to other
algorithms. This is because interference aware and nearest
neighbor algorithms always selects the neighbor which is
closer. Hence such algorithms are prone to have larger number
of intermediate hops resulting in high end-to-end delay and
incentives paid.
E. End-to-end throughput analysis
We have plotted minimum guaranteed end-to-end throughput by varying the number of nodes in Fig. 9. From Fig. 9
21st International Conference on
Advanced Information Networking and Applications Workshops (AINAW'07)
0-7695-2847-3/07 $20.00 © 2007
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on December 3, 2008 at 00:29 from IEEE Xplore. Restrictions apply.
0.45
1.05
Proposed algorithm
Optimum hop size routing
Interference aware routing
Nearest neighbor routing
Nearest neighbor algorithm
0.4
Intereference aware algorithm
1
Optimum hop size algorithm
Proposed algorithm
0.95
End−to−end throughput
End−to−end delay in sec
0.35
0.3
0.25
0.2
0.9
0.85
0.8
0.75
0.7
0.15
0.65
0.1
0.6
0.05
500
1000
Fig. 7.
1500
2000
Number of nodes
2500
3000
End-to-end delay versus number of nodes
Fig. 9.
12
11
Total incentives paid in units
10
Nearest neighbor algorithm
Interference aware algorithm
Optimum hop size algorithm
Proposed algorithm
9
8
7
6
5
4
3
2
500
Fig. 8.
1000
1500
2000
Number of nodes
2500
0.55
500
3000
Incentives spent per communication against number of nodes
we can deduce that the minimum guaranteed throughput in
the case of proposed algorithm is considerably higher. This
is because from (2) it is clear that end-to-end throughput is
function of number of hops. As the number of hops increases
the end-to-end throughput reduces.
VIII. C ONCLUSION
In our work, we have proposed a unified routing algorithm
for MCN by taking several QoS metrics into consideration.
The proposed algorithm has been compared with the existing MCN routing algorithms such as; interference aware
algorithm, nearest neighbor algorithm and optimum hop size
algorithm. We conclude that the proposed algorithm is superior
in terms of incentives paid, interference, total power spent,
call dropping, end-to-end throughput and end-to-end delay
compared to other algorithms.
R EFERENCES
[1] Y. Lin and Y. Hsu, “Multi-hop cellular: A new architecture for wireless
communications,” in Proc. IEEE INFOCOM 2000, vol. 3, 26-30 March
2000, pp. 1273–1282.
[2] I. Ioannidis and B. Carbunar, “Scalable routing in hybrid cellular and
ad-hoc networks,” in IEEE International Conference on Mobile Ad-hoc
and Sensor Systems, vol. 25-27 Oct, 2004, pp. 522–524.
1000
1500
2000
Number of nodes
2500
3000
End-to-end throughput versus number of users
[3] Y. H. Sreng. V and F. D, “Coverage enhancement through two-hop
relaying in cellular radio systems,” in Wireless Communications and
Networking Conference, 2002. WCNC2002, vol. 2, 17-21 March 2002,
pp. 881–885.
[4] C.-W. L. Jang-Ping Sheu and C.-M. Chao, “Power-aware routing for
energy conserving and balance in ad hoc networks,” in IEEE International Conference on Networking, Sensing and Control, 2004, vol. 1,
21-23 Mar 2004, pp. 468–473.
[5] G. Heijenk and F. Liu, “Interference-based routing in multi-hop wireless
infrastructures,” Eleseevier: Computer Communications, vol. 29, pp.
2693–2701, 2006.
[6] A. A. N. A. Kusuma, L. L. H. Andrew, and S. V. Hanly, “On routing
in cdma multihop cellular networks,” in Globecom 2004, 29 Nov-3 Dec
2004, pp. 3063–3067.
[7] E. M. Royer and C. K. Toh, “A review of current routing protocols
for ad-hoc mobile wireless networks,” IEEE Personal Commun. Mag.,
vol. 6, no. 2, pp. 46–55, Apr 1999.
[8] J.-P. H. Ben Salem, Levente Buttyan and M. Jajobsson, “Incentives
in practice: On the benefits and feasibility of incentive based routing
infrastructure,” in ACM SIGCOMM workshop on Practice and theory of
incentives in networked systems, September 2004.
[9] R. Janefalkar A.A, Josiam K, “Cellular ad-hoc relay for emergencies (care),” in IEEE 60th Vehicular Technology Conference, 2004.
VTC2004, vol. 4, 26-29 Sept 2004, pp. 2873–2877.
[10] Y. X. B. Li, “A location-aided power-aware routing protocol in mobile ad
hoc networks,” in IEEE Global Telecommunications Conference, 2001
GLOBECOM ’01, vol. 5, 25-29 Nov 2001, pp. 2837–2841.
[11] W. Liang and Y. Yuansheng, “Maximizing battery life routing in
wireless ad hoc networks,” in Proceedings of the 37th Annual Hawaii
International Conference on System Sciences, 2004, 5-8 Jan 2003, pp.
460–469.
[12] J.-H. Chang and L. Tassiulas, “Maximum lifetime routing in wireless
sensor networks,” IEEE/ACM Trans. Networking, vol. 12, no. 4, pp.
609–619, Aug. 2004.
[13] M. Haenggi, “On routing in random rayleigh fading networks,” IEEE
Trans. Wireless Commun., vol. 4, no. 4, pp. 1553–1562, Jul 2005.
[14] Y. Q. Peter Marbach, “Cooperation in wireless ad hoc networks: A
market-based approach,” IEEE/ACM Trans. Networking, vol. 13, pp.
1325–1338, Dec 2005.
[15] (2003) A charging and rewarding scheme for packet forwarding
in multi-hop cellular networks. [Online]. Available: http://www.
informatics.indiana.edu/markus/papers/routingpay2.pdf
[16] M. Zorzi and S. Pupolin, “Optimum transmission ranges in multihop
packet radio networks in the presence of fading,” IEEE Trans. Commun.,
vol. 43, no. 7, pp. 2201–2205, Jul 1995.
[17] F. Xue and P. Kumar, “The number of neighbors needed for connectivity
of wireless networks,” Springer:Wireless Networks, vol. 10, pp. 169–181,
Mar 2004.
[18] G. L. S. C. S. Patel C.S and T. G. Pratt, “Simulation of rayleigh-faded
mobile-to-mobile communication channels,” IEEE Trans. Commun.,
vol. 53, no. 11, pp. 1876–1884, Nov 2005.
21st International Conference on
Advanced Information Networking and Applications Workshops (AINAW'07)
0-7695-2847-3/07 $20.00 © 2007
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on December 3, 2008 at 00:29 from IEEE Xplore. Restrictions apply.