You are on page 1of 7

Spatial Packing of Nodes in a Wireless Ad-Hoc

Network With Long and Short Hops


Adnan Fazil and Aamir Hasan
Department of Avionics Engineering, Institute of Avionics
and Aeronautics (IAA), Air University, E-9, islamabad, Pakistan.
Email: {adnan.fazil, aamir.hasan}@mail.au.edu.pk

AbstractScheduling in a wireless ad hoc network is necessary for sharing of the wireless channel by different nodes
to communicate successfully i.e. while meeting the required
SINR threshold. Contrary to the infra-structure based wireless
networks, the wireless ad hoc network require a scheduling
algorithm that should be easily implementable with as little
as possible coordination between different nodes in distributed
fashion. In this paper, we have proposed a receiver guard zone
based scheduling scheme for a wireless ad hoc network with
long and short hops. Our proposed scheduling scheme has a
performance 74% as compared to centralized scheduling
scheme that is considered as optimal but practically infeasible
for wireless ad hoc networks. The proposed scheduling scheme
also has a remarkable gain in transmission capacity as compared
to random ALOHA and Carrier Sense Multiple Access (CSMA).

I. I NTRODUCTION
The need of an efficient ad hoc network is highly desirable
in todays world because of its properties which can be very
useful with new technologies, devices and applications. That
is why we have seen a lot of research and innovations in
this field since last few years, especially on the topic of
capacity and power control in wireless ad hoc networks [1]
[6]. The Medium Access Control (MAC) protocol plays a very
important role in design of a wireless ad hoc network and
has a major impact on capacity due to distributed nature of
this network and simultaneous transmission in same channel
by different contending nodes [3], [7]. Although, network
capacity is an important parameter to be considered while
designing an ad hoc network, but in practical networks, other
parameters like energy efficiency, quality of service, system
robustness, delays etc. should also be considered [8]. In this
paper we are focusing on capacity of a wireless ad hoc network
like in [1], [2], [9], [10].
A. Related Work
The requirement that a wireless communication channel has
to be shared by all the communication devices in a network,
highlights the importance of scheduling in wireless ad hoc
networks [3], [11], [12] and the implementation technique of
a scheduling scheme has major impact on some of network
parameters like end-to-end delay, outages, power efficiency,
data throughput and channel capacity. In [7] the authors used
a stochastic model to find out the optimal scheduling technique
for an ad hoc network. They propose a Medium Access
Control (MAC) protocol which ensures a random exclusion

zone around each node in the network that maximizes the


special reuse but at the cost of higher number of failed
transmissions i.e. greater than 50%. The work in [13] lacks
the analysis of the proposed scheduling scheme in terms of
its special reuse or the ability to protect existing links while
adding new links in the network.
A scheduling algorithm assuming global knowledge of
attenuation in all the transmissions and interference paths was
proposed in [2] along with a power control algorithm that
guaranties optimal power assignment for scheduled subset
of transmissions. Although the proposed scheme is optimal,
it has two obvious short comings. First, such a scheme is
impractical to implement due to absence of any centralized
controller and global information in a wireless ad hoc network,
although it provides a good performance comparison to other
practical algorithms which are suboptimal. Second and more
importantly, incorporating a tight power control algorithm
limits the ability of scheduling scheme to add additional links.
This would be the case when the existing links Signal to
Interference plus Noise Ratio (SINR) requirement cannot be
met as new links are admitted in the system or even if the
SINR requirement can be met, the power assignment of the
existing links would have to be recomputed.
The work in [15] implements a multistage contention protocol in a distributed fashion to realize the spatial packing
which also achieves good performance (close to optimal). But,
the model assumes fixed transmission distances without using
any power control method. In contrast to [9] where a simple
ALOHA type MAC is employed we propose a receiver guard
zone based scheduling scheme as in [3], [14], [16] but we
considered a network with long and short hops instead of
fixed Tx-Rx separation. The work in [3] proposed a system
wide fixed guard zone size, but here we propose a guard
zone that vary with the Tx-Rx separation of the interfering
transmitter i.e. each receiver has different guard zone for
different interfering transmitter and try to optimize the guard
zone radius to have a best compromise between spatial reuse
and probability of outage under Pairwise Power Control (PPC)
scheme [9]. We have considered a realistic network of finite
extent with uniformly distributed nodes in contrast to [17].
B. Main Result
In this paper we have studied the effect of spatial separation
between simultaneous transmissions in an ad hoc network.

TABLE I
N ETWORK PARAMETERS , UNLESS OTHERWISE SPECIFIED .

Symbol
R
n

dmin
N

Description
Network Radius
Receiver Noise
Path loss exponent
Minimum required SINR(SIN Rmin )
Short hop Tx-Rx separation
Number of nodes in the network

Value
50m
0.001
4
10dB
R/20
100

Spatial separation between simultaneous transmissions can be


enforced by introducing a guard zone around each active
receiver, where guard zone is defined as a region around
an active receiver where transmissions are prohibited by any
transmitter except the intended transmitter for that receiver.
In this paper we have proposed an optimal guard zone and
scheduling scheme which maximizes spatial reuse and can be
implemented easily in a distributed fashion. In the simulation
results we demonstrate that the proposed guard zone based
scheduling scheme gives us a huge improvement in capacity as compared to networks without scheduling i.e random
ALOHA networks. The proposed scheme also achieves good
capacity ( 74%) as compared to the well known optimal but
practically infeasible centralized scheduling scheme.
The rest of this paper is distributed as follows. Section II
explains the system model and also defines some assumptions
used in this study, section III explains the well known near
optimal scheduling algorithm in detail with its implementation
procedure in our simulations. Then section IV is dedicated
to explain the proposed receiver guard zone based scheduling scheme with power control scheme used in this paper,
section V explains the detailed analysis of the proposed
scheduling scheme when compared with random ALOHA,
CSMA and near optimal Joint Scheduling and Power Control
Algorithm (JSPCA). At the end the work is concluded in
section VI.
II. S YSTEM M ODEL AND A SSUMPTIONS
The system model used in this paper utilizes a homogeneous
Poisson Point Process (PPP) for generation of transmitters in
a circular region of radius R meters with density i nodes per
meter square. The receiver for every transmitter is placed at
a distance of dmin or dmax , with 50% probability for each
to generate the short hop and long hop Tx-Rx separations, at
an angle that is uniformly distributed from 0 to 2 radians.
Therefore each transmitter has its own intended receiver.
Stationary Poisson Point Process is a reasonable assumption
for generating nodes in an ad-hoc network [3]. This paper
considers single hop transmissions analysis only, as we are
viewing the network at single time snapshot and all the
nodes are considered to have an omni directional antenna.
Capacity or transmission capacity is defined in this thesis as
the maximum possible density of simultaneous transmissions
that satisfies a minimum target SINR (SIN Rmin = ) at

each receiver. In this paper, we have assumed a simple path


loss model for signal propagation, and neglected routing, endto-end delay, and energy efficiency. Other network parameters
are as given in table I.
III. J OINT S CHEDULING AND P OWER C ONTROL
A LGORITHM (JSPCA)
Optimum power control under a maximum power constraint
for channelized cellular systems which maximizes spatial
reuse was presented in [18], [19]. The results by these work
were proved by [2] to be also applicable in wireless ad hoc
networks. A JSPCA of [2] limits the interference generated by
simultaneous transmissions in order to increase spatial reuse in
wireless ad hoc networks. They used the scheduling algorithm
which assumes global knowledge of attenuation in all the
transmissions and interference paths to determine the largest
subset of simultaneous transmitters. Their power control algorithm guaranties optimal power assignment for scheduled
subset of transmissions by using foschinis algorithm [19]. The
mathematical description and implementation of this scheme
is given below.
The Joint Scheduling and Power Control Algorithm
(JSPCA) is being implemented using [2], [19]. Starting with
the initial PPP distribution, arrange the pairs in a random
sequence which defines their priority of communication i.e. the
communication between which Tx-Rx pair is most important.
Start with the first one and analyze using [19], whether an
optimal power vector exists or not, for the very first Tx-Rx pair
the power vector will always exist as there are no interferers
for the first pair. Then include the second pair and analyze that
whether an optimal power vector exists for both transmitters
or not. If the vector exists then keep both of them in the
network and if it doesnt exist then discard the second pair (i.e.
dont allow one with the least priority to communicate) and
analyze again after including the next Tx-Rx pair for existence
of optimal power vector. This procedure is same as [2].
Lets we have total N initial Tx-Rx pairs in the network
with each having a minimum SINR requirement of SIN Rmin
denoted by . JSPCA will be applied on the initial distribution of nodes ( i.e. PPP distribution) with nodes density i .
Expression for SINR on lth receiver under JSPCA is as given
in (1)
l =

n+

G P
P ll l
i, l {1, 2, ..., N }
i6=l Gli Pi

(1)

Where,
Gli = Gain from ith transmitter to lth receiver i.e. equal to
1
(dli ) , where dli is the distance between ith transmitter and
lth receiver with free space path loss exponent .
Pi = Power transmitted by ith transmitter.
n = Noise power (thermal noise), considered equal on all
the receivers.
The condition i 6= l is necessary because the power
transmitted by ith transmitter is considered as interference
for lth receiver but this is not the case when i = l because

then the power transmitted by ith transmitter is the actual


communication signal for the lth receiver.

dll
, i 6= l
dli
Fli = 0, i = l
Fli =

(2a)
(2b)

Where, = Minimum SINR required at all receivers and F


is an irreducible matrix with non-negative elements as defined
in (2a) and (2b) . The optimal power vector exists if the perronfrobineous (maximum modulus) Eigen value of matrix F is
less than unity [18], [19]. We will start with first Tx-Rx pair
and keep on analyzing this matrix F by including the pairs
from initial distribution one by one, any new included pair
which gives the perron-frobineous Eigen value greater than or
equal to unity will be excluded from the initial distribution.
In this way, after analyzing the complete distribution, we will
get a scheduled distribution at the end and that will have an
optimal power vector PP areto . The scheduled nodes density
that will be less than or equal to the initial nodes density,
due to exclusion of nodes during scheduling, is denoted by
P areto i .
IV. T HE P ROPOSED A PPROACH
JSPCA is the optimal scheduling algorithm but it is not
feasible for wireless ad hoc networks because of two major reasons discussed earlier. We propose here a distributed
scheduling algorithm for wireless ad hoc networks which is
based on receiver guard zone scheduling and Pairwise Power
Control (PPC).
A. Guard Zone Based Scheduling
In a random access protocol such as random ALOHA, all
the nodes transmit without contending with each other to
have free transmission medium. Naturally when the nodes
density is high in the network, the transmissions under random
ALOHA results in very high potability of outage due to
excessive interference that may not be acceptable for successful communication link between two nodes. One way to
suppress interference is by ensuring spatial separation among
concurrent transmissions by incorporating a guard zone around
active receivers where transmitters (other than the intended
transmitter) are inhibited [3], [16]. The guard zone is defined
as the region around an active receiver where no interfering
transmissions are allowed. There is a trade off between trying
to have minimum probability of outage at each receiver (i.e.
using largest possible guard zone size) and maximizing the
number of simultaneous transmissions in the network (i.e using
smallest possible guard zone size).
The scheduling in wireless ad hoc networks is a very diverse
and rich research topic but a very simple way to find out
the necessity and benefits of scheduling in ad hoc networks
is to use optimal guard zone scheduling [3], [14].In contrast
to [9] where a simple ALOHA-type MAC is employed,
an optimal guard zone of size DOpt around each receiver
is implemented that helps limit the aggregate interference

by inhibiting the nearby dominant interferers and increase


the average probability of successful communication at each
receiver. where optimal guard zone radius around each receiver
is defined as DOpt = (1 + Opt )dtx , where opt is a variable
which needs to be optimized to have maximum transmission
capacity and dtx is the Tx-Rx separation of the interfering
transmitter, therefore dtx = dmin or dmax . This means that a
receiver inhibits an interfering transmitter of Tx-Rx separation
dmin if it is placed at a distance of less than or equal to
(1 + Opt )dmin and inhibits an interfering transmitter of TxRx separation dmax if it is placed at a distance of less than
or equal to (1 + Opt )dmax .
The guard zone-based scheduling algorithm selects a subset
of initial nodes which are allowed to communicate simultaneously. The algorithm selects this subset of transmitters based
solely on the guard zone criteria from the initial transmission
scenario of N contending Tx-Rx pairs with initial nodes
density of i . The optimal value of guard zone corresponds
to the optimal value of variable i.e. Opt . The variable
varies from 1 to where = 1 is the case where there is
no guard zone or no scheduling i.e. random ALOHA. Guard
zone based scheduling gives us a subset of initial nodes with
nodes density GZ where GZ i .
B. Pairwise Power Control and Probability of Outage
The pairwise power control is a technique in which each
transmitter sets its transmission power such that the communication signal power received at its intended receiver is a
constant i.e. . Therefore according to this definition, if a
transmitter and receiver are separated by a distance d then
the transmitter should have a transmission power equal to
Pt = (d) so that the received signal power at the receiver
is equal to after signal attenuation.This technique is called
pairwise power control because every transmitter and receiver
pair chooses the transmission power without knowing anything
about other Tx-Rx pairs in the network [9]. Therefore each
transmitter in this network will have a transmission power
of P1 or P2 depending on its Tx-Rx separation distance
as P1 = (dmin ) and P2 = (dmax ) . The density of
transmitters that are transmitting simultaneously under PPC is
GZ but the density of receivers with received SINR greater
than or equal to is GZppc which might be less than GZ
due to outages under PPC. Therefore the probability that a
scheduled receiver will be successfully communicating under
PPC is given by
P [SIN Ri ] = 1
GZ GZppc
GZ

(3)

is the probability of outage after apwhere =


plying pairwise power control (PPC) on guard zone scheduled
nodes and GZppc = GZ (1 ) is the transmission capacity
of the network under guard zone based scheduling and PPC i.e.
the proposed scheme. To find the optimal guard zone radius,
we plot network capacity versus where maxima of the
network capacity gives us = Opt as shown in figure 1
where Opt = 0.9.

A. Proposed Scheme Versus Random ALOHA


The random ALOHA scheme is similar to guard zone
scheduling at = 1 which is equal to guard zone based
scheduling with zero guard zone radius or no scheduling
at all. The transmission capacity under proposed scheduling
scheme is compared with transmission capacity of random
ALOHA in this section. The gain in transmission capacity
under proposed scheme as compared to random ALOHA is
denoted by GALOHA and defined in (4).
GALOHA =

GZppc
ALOHA

(4)

Fig. 1. Initial Nodes density (i ), density of nodes after only guard zone
based scheduling (GZ ), Transmission capacity under JSPCA (P areto )
and transmission capacity under proposed scheme (GZppc ) versus . The
maxima of GZppc at = 0.9 shows that the optimal value of guard zone
correspond to Opt = 0.9 for this network at these network parameters.

V. S IMULATION R ESULTS
The optimal value of guard zone for this network was
calculated as described in previous section for different network parameters at N = 100 and then used for simulations
at different values of N . The network transmission capacity
under proposed scheme is then analyzed by comparison with
Transmission capacity of JSPCA, Transmission capacity of
CSMA which uses guard zone around transmitters and Transmission capacity of ALOHA which doesnt use any type of
scheduling.

(a) At different values of .

(b) At different values of dmax .

Fig. 2. Transmission capacity under JSPCA (P areto ), transmission capacity


under CSMA, transmission capacity under random ALOHA and transmission
capacity under proposed scheme (GZppc ) versus initial number of contending trasmitters in the network (N ).

The transmission capacities under all these four schemes


versus N is given in figure 2 which shows the benefits of
proposed scheme up to some extent. The detailed comparison
of proposed scheme with other three schemes is given in
following subsections.

Fig. 3. (a) GALOHA is plotted for dmax = 2dmin at = 4 and = 5 to


analyze effect of path loss exponent on GALOHA . (b) GALOHA is plotted
for dmax = dmin and dmax = 2 dmin at = 4 to analyze effect of long
hop Tx-Rx separation on GALOHA .

Figure 3(a) shows plots of GALOHA at = 4 and = 5


for a network with long and short hops, where long hop
is twice the short hop i.e. dmax = 2 dmin . As we can
see from the plots when number of initial nodes in the
network (N ) is small, the majority of competing nodes are
able to transmit simultaneously with successful reception at
their respective receivers because of the inherent separation in

initial nodes Poisson point distribution and the performance


of random ALOHA is not that bad. The proposed guard zone
based scheduling drastically outperforms random ALOHA at
large N because it utilizes space more efficiently. We can
also see that for small value of , the value of GALOHA
reaches above 400 because small path loss exponent rewards
scheduling. Figure 3(b) shows plots of GALOHA for two
networks , one with all the Tx-Rx separations equal to 2.5m
(i.e. dmax = dmin ) and other with long and short hops
where dmax = 2 dmin . This figure shows that the proposed
scheduling is much more beneficial for long and short hop
networks where its gain compared to random ALOHA exceeds
400.
B. Proposed Scheme Versus CSMA
The MAC protocol in IEEE 802.11 uses Carrier Sense
Multiple Access (CSMA) scheme for simultaneous transmissions by different nodes in a wireless network. Since CSMA
is the technique which is currently used by Wireless Local
Area Network (WLAN), therefore the comparison of proposed
guard zone based scheduling with CSMA will be a useful
analysis. CSMA protocol is equivalent to have a system wide
fixed guard zone around each active transmitter instead of
receiver. For comparison, we simulated the optimal CSMA by
optimizing the spatial reuse and consequently the transmission
capacity for the network used in this paper.
The results for GCSM A which is defined as the gain
in transmission capacity under proposed guard zone based
scheduling as compared to transmission capacity under CSMA
GZppc
are shown in figure 4. As figure shows
i.e. GCSM A = CSM
A
clearly, the proposed scheduling performs better than CSMA
specially when N (initial nodes density) is large. Figure 4(a)
shows that the proposed scheduling gives us more than 50%
gain in capacity at large N . We can also see that small value of
path loss exponent () rewards proposed scheduling for large
initial nodes density (i.e. at large N ), but GCSM A has large
value for large at small N . This also shows that GCSM A is
less sensitive to path loss exponent than GALOHA .
Figure 4(b) shows that the proposed scheduling has much
more benefit for a network with long and short hops as it
can accommodate long hops more easily as compared to
CSMA. We can see that if our network has a constant Tx-Rx
separation throughout the network, then the proposed guard
zone scheduling performs almost 23% better than CSMA, but
if the network has a combination of short and long hops then
our proposed scheduling outperforms the widely used CSMA
by more than 50% .
C. Proposed Scheme Versus JSPCA
A global search with central scheduler for an ad hoc
network was proposed in [2] to maximize spatial reuse, which
determines the largest subset of the initial contending nodes
that can communicate simultaneously. Randomly ordering the
Tx-Rx pairs from 1 to N in this paper is definitely suboptimal
in the sense that the scheduling phase schedules transmissions
in a sequence, instead of searching for the largest possible

(a) At different values of .

(b) At different values of dmax .


Fig. 4. (a) GCSM A is plotted for dmax = 2 dmin at = 3 and = 4
to analyze effect of path loss exponent on GCSM A . (b) GCSM A is plotted
for dmax = dmin and dmax = 2 dmin at = 4 to analyze effect of long
hop Tx-Rx separation on GCSM A .

subset of transmitters which satisfy the JSPCA or guard zone


criteria. But this is necessary for wireless ad hoc networks,
because of their distributed nature, since the size of the search
space grows exponentially with the initial nodes density or
N . The selection of transmitters from a random sequence
one by one is similar to a realistic implementation of ad
hoc network e.g. scheduling the transmitters in a sequence
of desired data/communication signal generation.
The results for transmission capacity gain with respect to
JSPCA which is defined as gain in transmission capacity
under proposed guard zone based scheduling as compared to

transmission capacity under JSPCA i.e. GP areto = PGZppc


areto
are shown in figure 5. We can see from figure 5(a) that the
proposed scheme gives us transmission capacity 74% of
the near optimal scheme (JSPCA). the gain i.e. GP areto starts
to decrease with N but then it starts to increase with N at
higher initial nodes density when transmission capacity under
JSPCA starts to saturate. we can also see that for a small
fixed number of initial contending nodes N , this gain increases
with increase in path loss exponent. But for a large fixed

(a) At different values of .

and CSMA for a wireless ad hoc network with long and short
hops and it can even outperform the near optimal JSPCA for
small path loss exponent at very large initial nodes density
under the assumed system model and network parameters.
The proposed scheduling and its simulated results are suboptimal because of some reasons (i) it searches in a sequence
after arranging all the nodes in a random manner instead of
searching for a largest possible subset (ii) it uses same guard
zone size around all receivers to inhibit a specific interferer
(iii) we optimized the guard zone radius for N = 100 and
then used that Opt for N = 1 300 for simulations, but
the performs can be even better than the simulated one if we
optimize guard zone size for each value of N i.e. if we can
define optimal guard zone size as a function of initial number
of contending nodes Opt = f (N ) which we have left for
future work. Now we can say along with these results that the
proposed guard zone based scheduling scheme for a wireless
ad hoc network at = 3 5 has a performance 74% of
near optimal but practically infeasible JSPCA.
R EFERENCES

(b) At different values of dmax .


Fig. 5. (a) GP areto is plotted for dmax = 2 dmin at = 3, 4 and 5 to
analyze effect of path loss exponent on GP areto . (b) GP areto is plotted for
dmax = dmin , dmax = 2 dmin and dmax = 3 dmin at = 4 to
analyze effect of long hop Tx-Rx separation on GP areto .

value of N , GP areto increases with decrease in path loss


exponent. We can also see that the proposed guard zone based
scheduling outperforms the near optimal JSPCA at very large
initial density and small value of path loss exponent.
Figure 5(b) shows the effect of long hop Tx-Rx separation
value on GP areto . We can see that for a small fixed value
of N , GP areto for a network with fixed Tx-Rx separation of
dmin is much better than GP areto for a network with long
and short hops. But for a very large fixed number of initial
nodes N , GP areto for a network with long and short hops is
much better than the other two networks, this is because of the
nature of JSPCA which prefers short hops while scheduling.
VI. C ONCLUSIONS
In this paper we have studied the effects of guard zone
based scheduling on a network with long and short hops.
We have seen that the proposed guard zone based scheduling
scheme used a guard zone to prohibit a transmitter depending
on interferers Tx-Rx separation. This study shows that the
proposed scheduling can perform much better than ALOHA

[1] P. Gupta and P. R. Kumaar, The Capacity of Wireless Networks, IEEE


Transaction on Information Theory, vol.46, pp. 388-404, March. 2000.
[2] T. ElBatt and A. Ephremides, Joint Scheduling and Power Control for
Wireless Ad hoc Networks, IEEE Transaction on Wireless Communications, vol. 3, no. 1, pp. 7485, Jan. 2004.
[3] A. Hasan and J. G. Andrews, The Guard Zone in Wireless Ad Hoc
Networks, IEEE Transaction on Wireless Communications, vol. 3, no.
3, pp. 897-06, Mar, 2007.
[4] Massimo Franceschetti, Olivier Dousse, David N. C. Tse and Patrick
Thiran, Closing the Gap in the Capacity of Wireless Networks Via
Percolation Theory, IEEE Transaction on Inform Theory, vol. 53, NO.
3, Mar 2007.
[5] S. Lv, W. Zhuang, X. Wang, and X. Zhou, Scheduling in wireless
ad hoc networks with successive interference cancellation, in Proc. Of
IEEE INFOCOM11, 2011.
[6] Han-Chiuan Luo, Eric Hsiao-Kuang Wu and Gen-Huey Chen, Minimizing Ceased Areas with Power Control for Spatial Reuse in IEEE 802.11
Ad Hoc Networks, IEEE 17th International Conference on Parallel and
Distributed Systems, 2011.
[7] F. Baccelli, B. Blaszczyszyn, and P. Muhlethaler, A Spatial reuse
ALOHA MAC Protocol for Multihop Wireless Mobile Networks,
Technical Report 4955, INRIA, Oct. 2003.
[8] M. Chiang and M. Yang, Towards network X-ities from a topological
point of view: Evolvability and scalability, in Proceedings of Allerton
Conf. on Comm., Control, and Computing, Sept. 2004.
[9] S. Weber, X. Yang, J. G. Andrews, and G. de Veciana, Transmission
Capacity of Ad Hoc Networks with Outage Constraint, IEEE Transaction on Information Theory, vol. 51, no. 12, pp. 4091-102, Dec. 2005.
[10] S. Toumpis and A. J. Goldsmith, Capacity Regions for Wireless Ad
hoc Networks, IEEE Transaction on Wireless Communications, vol. 2,
no. 4, pp. 73648, July 2003.
[11] S. Ramanathan and E. L. Lloyd, Scheduling Algorithms for Multihop
Radio Networks, IEEE/ACM Transaction Net., vol. 1, no. 2, pp. 16677,
Apr. 1993.
[12] A. Muqattash, M. Krunz, and W. E. Ryan, Solving the Near-far Problem
in CDMA-based Ad hoc Networks, Ad hoc Networks Journal, vol. 1,
no. 4, pp. 43553, Nov. 2003.
[13] A. Muqattash and M. Krunz, CDMA-based MAC protocol for wireless
ad hoc networks, in Proc. SIGMOBILE, June 2003, pp. 153164.
[14] A. Hasan and J. G. Andrews. Scheduling using near-optimal guard
zones for CDMA ad hoc networks, In Proc. IEEE International
Conference on Communications, volume 9, pages 40024007, Jun. 2006.
[15] X. Yang, A. Hasan, G. de Veciana, and J. G. Andrews, MAC Protocols
for Spread Spectrum Ad Hoc Networks: Spatial Thinning vs. Packing,
in Proc. CISS, Mar. 2005.

[16] Aamir Hasan, M. Ali and Zahid Ali, Modeling Aggregate Interference
in Ad Hoc Networks, International Bhurban Conference on Applied
Sciences & Technology (IBCAST), January, 2010.
[17] S. Weber, J. G. Andrews, and N. Jindal, An overview of the transmission capacity of wireless networks, IEEE Transaction on Communications, vol. 58, pp. 35933604, December 2010.
[18] Jens Zander, Performance of Optimum Transmitter Power Control
in Cellular Radio Systems, IEEE TRANSACTIONS ON VEHICULAR
TECHNOLOGY, VOL. 41, NO. 1, FEBRUARY 1992.
[19] G. J. Foschini and Z. Miljanic, A Simple Distributed Autonomous
Power Control Algorithm and its Convergence, IEEE TRANSACTIONS
ON VEHICULAR TECHNOLOGY, vol. 43, no. 4, November. 1993.