Sunteți pe pagina 1din 10

20 (IJCNS) International Journal of Computer and Network Security,

Vol. 2, No. 6, June 2010

Power Management in Wireless ad-hoc Networks


with Directional Antennas
Mahasweta Sarkar1 and Ankur Gupta2
1
Department of Electrical and Computer Engineering
San Diego State University, San Diego, California, USA
Email: msarkar2@mail.sdsu.edu
2
Department of Electrical and Computer Engineering
San Diego State University, San Diego, California, USA
Email: ankur275@gmail.com

ad-hoc sensor or hybrid ad-hoc network consists of a


Abstract: Power control in wireless ad-hoc networks has been number of sensor spreads in a geographical area. Each
known to significantly increase network lifetime and also impact sensor is capable of mobile communication and has some
system throughput positively. In this paper, we investigate the level of intelligence to process signals and to transmit data.
virtues of power control in an ad-hoc network where nodes are In order to support routed communications between two
equipped with directional antennas. We use a cross-layer power mobile nodes, the routing protocol determines the node
optimization scheme. The novelty of our work lies in the fact connectivity and routes packets accordingly. This makes a
that we exploit the knowledge of the network topology to
mobile ad-hoc sensor network highly adaptable so that it can
dynamically change our power control scheme. The information
about the network topology is retrieved from the Network Layer.
be deployed in almost all environments like military
Specifically, we consider three different network topologies – (i) operations, security in shopping malls and hotels or to locate
a sparse topology, (ii) a dense topology and a (iii) cluster-based a vacant parking spot.
topology. The MAC layer, implements a dynamic power control Each node in the network is equipped with an antenna
scheme which regulates the transmission power of a (omni directional or directional) which enables it to find its
transmitting node based on the node density at the receiver
neighbors and communicate with them. Since the antenna
node. In addition a transmitting node can adjust its
transmission power to the optimal level if the SINR value at the
plays such an important role in the communication, the
receiving node is known. Our cross-layer power control choice of the antenna becomes critical. With the existing
algorithm takes the above fact into account. Extensive schemes, an omni directional antenna was used both at the
simulation in NS2 shows that in all the three topologies our transmitter and at the receiver end. Although using this
cross layer design (use of directional antennas with power configuration does enable us to achieve the task of
control) leads to prolonged network lifetime and significantly communication, there are many drawbacks of using an omni
increases the system throughput. direction antenna as compared to using a directional one.
Extensive research in MAC and PHY layer protocols is
Keywords: ad hoc network, power management, being conducted to support these antennas.With the
directional antenna, cross layer advancement in antenna technology it has become possible
to use array antennas and purely directional antennas in ad-
1. Introduction hoc networks. Some of the advantages of using directional
antennas as compared to omni directional antennas are
An ad-hoc or short-live network is the network of two or increased range, spatial reuse and reduced collisions. Since
more mobile devices connected to each other without the the antenna beam can be pointed to a particular direction
help of intervening infrastructure. In contrast to a fixed and all the energy can be focused there, the range is
wireless network, an ad-hoc network can be deployed in increased. Unlike omni directional antennas, by using
remote geographical locations and requires minimum setup directional antennas we can have more nodes
and administration costs. Moreover, the integration of an communicating in the same space, thereby exploiting spatial
ad-hoc network with a bigger network-such as the Internet- reuse and increasing the throughput of the system. Since
or a wireless infrastructure network increases the coverage omni directional antennas transmit and receive in all the
area and application domain of the ad-hoc network. directions, they are more prone to collisions, which is not
However, communication in an ad-hoc network between the case when using directional antennas. Now that the
different hosts that are not directly linked is an issue not choice of antennas has been discussed, we discuss the main
only for search and rescue operations, but also for focus of this paper which is power control. The first
educational and business purposes. drawback of using the fixed-power approach is that it
negatively impacts channel utilization by not allowing
An ad-hoc network can be classified into two main types:
concurrent transmissions to take place over the reserved
mobile ad-hoc network and mobile ad-hoc sensors network.
floor. The second drawback of the fixed-power approach is
Unlike typical sensor networks, which communicate directly
that the received power may be far more than necessary to
with the centralized controller, a mobile ad-hoc sensor
achieve the required signal-to-interference-and-noise ratio
network follows a broader sequence of operational scenarios,
(SINR), thus wasting the node’s energy and shortening its
thus demanding a less complex setup procedure. A mobile
lifetime. Therefore, there is a need for a solution, possibly a
(IJCNS) International Journal of Computer and Network Security, 21
Vol. 2, No. 6, June 2010

cross layer one that allows concurrent transmissions to take In this paper we start by looking at the power management
place in the same vicinity and simultaneously conserves schemes used in ad-hoc networks in section II. We then
energy. present our cross layer design in section III. Simulation
setup in section IV is followed by results and analysis in the
Ad-hoc networks could be broadly classified into three Vth section. We conclude this paper in the VIth section
topologies namely, spare, dense and cluster form. Since the followed by the references in the VIIth section.
MAC and Network layers play an important role in
conserving power and increasing the system throughput, it 2. Previous Work
is essential to have a cross layer design that can support
different topologies. The MAC layer in a system controls The research community has done a lot of work to suggest
how and when a node accesses the medium. This also potential ways of power control in ad-hoc networks. In [1]
includes the transmission power used for communication. If Rong Zheng and Robin Kravets proposed an on demand
a transmitting node is aware of the ideal SINR value at the power management scheme for ad-hoc networks. It is an
receiver, it can adjust its transmitting power such that only extensible on-demand power management framework for
required amount of power is used for communication. In ad-hoc networks that adapts to traffic load. Nodes maintain
doing so not only does the transmitter save power (compared soft-state timers that determine power management
to transmitting at full power) but also interferes with lesser transitions. By monitoring routing control messages and
nodes (as increased power would mean increased range). In data transmission, these timers are set and refreshed on-
case of sparse network, we introduce the “locally aware” demand. Nodes that are not involved in data delivery may
power management protocol in which a receiver can inform go to sleep as supported by the MAC protocol. This soft
its transmitter of the minimum required SINR (to be able to state is aggregated across multiple flows and its
decode the signal) so that the transmitter can adjust its maintenance requires no additional out-of-band messages.
transmitting power accordingly. In a dense topology a The implementation is a prototype of the framework in the
receiver may face interference from more nodes than in a NS-2 simulator that uses the IEEE 802.11 MAC protocol.
sparse topology. We propose the “globally aware” protocol Simulation studies using this scheme with the Dynamic
here. In such a case it is essential that the receiver transmits Source Routing protocol show a reduction in energy
its minimum required SINR value in all the directions. This consumption near 50% when compared to a network
will ensure that all the transmitters in this receiver’s without power management under both long-lived CBR
neighborhood will control their power such that they do not traffic and on-off traffic loads, with comparable throughput
interfere with this receiver anymore but can still carry on and latency. Preliminary results also show that it
with their own communications. Both the sparse and the outperforms existing routing backbone election approaches.
dense topology designs are supported by a power aware In [2] Zongpeng Li and Baochun Li present a probabilistic
routing protocol at the Network layer. Since a sensor power management scheme for ad-hoc networks. It
network is a network in which nodes gather relevant data introduces Odds, an integrated set of energy-efficient and
about an activity and send it to an actor node which fully distributed algorithms for power management in
processes the data further; it is important that efficient wireless ad-hoc networks. Odds is built on the observation
routing is performed from the sensor nodes to the actor that explicit and periodic re-computation of the backbone
nodes. Cluster topologies are very common in ad-hoc topology is costly with respect to its additional bandwidth
sensor networks to record signals of interest at certain places overhead, especially when nodes are densely populated or
only. Fig. 1 shows the difference between a homogeneous highly mobile. Building on a fully probabilistic approach,
and non-homogeneous spatial distribution of nodes. We Odds seek to make a minimum overhead, perfectly
study the changes in power consumed and throughput when balanced, and fully localized decision on each node with
all cluster nodes transmit at the same power to when there is respect to when and how long it needs to enter standby
power control. Power control is implemented by making mode to conserve energy. Such a decision does not rely on
sure that each cluster uses its own power level such that it periodic message broadcasts in the local neighborhood, so
does not interfere with other cluster nodes. This simple cross that Odds are scalable as node density increases. Detailed
layer design helps prolong the network lifetime and also mathematical analysis, discussions and simulation results
shows increase in throughput. have shown that Odds are indeed able to achieve our
objectives while operating in a wide range of density and
traffic loads. Power control schemes using purely directional
transmission and reception have not been researched in
detail. Exploiting the advantages of directional antennas
[13, 14, 15, 20] shows considerable improvement in the
overall network performance.

MAC layer solutions have been proposed from a long time.


In [3] a scheme for controlling the transmission power is
presented. More specifically to the effects of using different
Figure 1. Difference between a Homogeneous and Cluster transmit powers on the average power consumption and
Form Spatial Distribution of nodes end-to-end network throughput in a wireless ad-hoc
environment. This power management approach reduces the
22 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 6, June 2010

system power consumption and thereby prolongs the battery control when nodes are non-homogeneously dispersed in
life of mobile nodes. Furthermore, the invention improves space. In such situations, one seeks to employ per packet
the end-to-end network throughput as compared to other ad- power control depending on the source and destination of
hoc networks in which all mobile nodes use the same the packet. This gives rise to a joint problem which involves
transmit power. The improvement is due to the achievement not only power control but also clustering. Three solutions
of a tradeoff between minimizing interference ranges, for joint clustering and power control are presented. The
reduction in the average number of hops to reach a first protocol, CLUSTERPOW, aims to increase the network
destination, reducing the probability of having isolated capacity by increasing spatial reuse. The second, Tunnelled
clusters, and reducing the average number of transmissions CLUSTERPOW, allows a finer optimization by using
including retransmissions due to collisions. The present encapsulation. The last, MINPOW, whose basic idea is not
invention provides a network with better end-to-end new, provides an optimal routing solution with respect to
throughput performance and lowers the transmit power. [6] the total power consumed in communication. The
and [8] also talk about transmission power control. contribution includes a clean implementation of MINPOW
at the network layer without any physical layer support. All
Topology control [4], [9] schemes have also been proposed. three protocols ensure that packets ultimately reach their
In [4] authors present Span, a power saving technique for intended destinations.
multi-hop ad-hoc wireless networks that reduces energy Power aware routing protocols such as Power Aware AODV
consumption without significantly diminishing the capacity [10] also contribute to conserving the network power.
or connectivity of the network. Span builds on the Changes have been suggested to the NS2 routing structure
observation that when a region of a shared-channel wireless to accommodate for a power aware routing protocol that is
network has a sufficient density of nodes, only a small aimed at increasing the network lifetime. [17, 18, 19] also
number of them need be on at any time to forward traffic for propose energy efficient schemes for wireless networks.
active connections. Span is a distributed, randomized
algorithm where nodes make local decisions on whether to 3. Proposed Scheme
sleep, or to join a forwarding backbone as a coordinator.
Each node bases its decision on an estimate of how many of 3.1 Power Control: The MAC Perspective
its neighbors will benefit from it being awake and the
amount of energy available to it. Authors use a randomized In this paper we propose a cross layer design in which the
algorithm where coordinators rotate with time, MAC layer and the Network layer functions can be modified
demonstrating how localized node decisions lead to a to achieve lesser power consumption and increase
connected, capacity-preserving global topology. throughput. The system mimics the 802.11 standard and
Improvement in system lifetime due to Span increases as the uses directional antennas at the physical layer. Since
ratio of idle-to-sleep energy consumption increases, and different situations demand different control mechanisms,
increases as the density of the network increases. the solutions in this paper are aimed at the three most
Simulations show that with a practical energy model, system common topologies in wireless ad-hoc networks, namely
lifetime of an 802.11 network in power saving mode with sparse, dense and cluster forms. A power controlled MAC
Span is a factor of two better than without. Span integrates protocol decides the power level for transmission for a
nicely with 802.11. When run in conjunction with the particular node. It is the power used to access the channel
802.11 power saving mode, Span improves communication and also to carry on subsequent transmissions. Researchers
latency, capacity, and system lifetime. A cone-based solution have used different metrics to determine the optimum
that guarantees network connectivity was proposed in [9]. transmission power; the SINR of the receiving node being
Each node i gradually increases its transmission power until the most popular [add ref]. We assume that there is a
it finds at least one neighbor in every cone of angle. Node i mechanism like GPS that enables the nodes to be aware of
starts the algorithm by broadcasting a “Hello” message at the network topology. This enables the receivers to calculate
low transmission power and collecting replies. It gradually the SINR level based on the number of transmitters and
increases the transmission power to discover more neighbors their distance from this receiver. Not only does this prolong
and continuously caches the direction in which replies are network lifetime but also allows more nodes to communicate
received. It then checks whether each cone of angle contains at the same time. In the proposed protocols the collision
a node. The protocol assumes the availability of directional avoidance information does not prevent interfering nodes
information (angle-of-arrival), which requires extra from accessing the channel. Instead, it bounds the
hardware. This scheme does not seem to work for transmission powers of future packets generated by these
directional antennas [16] as most of the replies would be lost nodes. This is unlike the RTS/CTS packets used in the
depending on where the source’s antenna is pointing. Some 802.11 scheme.
researchers proposed the use of a synchronized global
signaling channel to build a global network topology To understand what this collision avoidance information is
database, where each node communicates only with its and how nodes can make use of it, consider the transmission
nearest N neighbors (N is a design parameter). This of a packet from some node i to some node j. Let SINR (i,j)
approach, however, requires a signaling channel in which be the SINR at node j for the desired signal from node i.
each node is assigned a dedicated slot. Then,
A scheme based on clustering [5] in ad-hoc networks was
also presented. This paper discusses the problem of power
(IJCNS) International Journal of Computer and Network Security, 23
Vol. 2, No. 6, June 2010

the CTS frame. On receiving the CTS frame, the transmitter


adjusts its power to the new level as specified by the CTS
(1) frame and continues further communication. Incase the
receiver needs to update the transmitter of any changes in
where, the SINR level, it can do so by including the new power
P (i,j)= received power at node j for a transmission from level information in any CTS frame which it would send out
node i ∑P (k,j)= power received as noise at j from its in response to a data frame. Since this protocol tries to
neighbors control the power levels of node pairs, we call it local power
hj= thermal noise (default value in NS2 is 101 dBm for the control. Also, since the nodes are sufficiently apart, the
802.11 std.) at node j. receiver does not need to inform all the nodes in the network
of its new SINR level.
A packet is correctly received if the SINR is above a certain
threshold (say, SINRth) that reflects the QoS of the link. By 2) Dense Networks: A Globally Aware Protocol
allowing nearby nodes to transmit concurrently, the
interference power at receiver j increases, and so SINR (i,j) The protocol which we have simulated states that each node
decreases. Therefore, to be able to correctly receive the in the network transmits with optimum power (which is the
intended packet at node j, the transmission power at node i minimum required SINR) in an attempt to maximize the
must be computed while taking into account potential future throughput of the network. When a transmitting node sends
transmissions in the neighborhood of receiver j. This is an RTS to the receiver successfully, the receiver piggy backs
achieved by incorporating an interference margin in the the SINR information in the CTS frame which can be
computation of SINR (i,j). This margin represents the further used to decide the ideal transmitting power at the
additional interference power that receiver j can tolerate transmitter. It also broadcasts the Collision Avoidance
while ensuring coherent reception of the upcoming packet Information in all the directions. When the neighboring
from node i. Nodes at some interfering distance from j can nodes receive this information, they adjust their
now start new transmissions while the transmission i → j is transmission power such that they can still carry on their
taking place. The interference margin is incorporated by conversation and not disrupt the other nodes. When a pair
scaling up the transmission power at node i beyond what is of communicating nodes experience interference from some
minimally needed to overcome the current interference at other node (SINR goes down at the reciever), the Collision
node j. For our simulations, a SINR margin of 10 dB Avoidance information is updated and can be sent to the
ensures correct reception. Also, it is to be kept in mind that transmitter with a CTS (in response to a data frame). This
the minimum transmission power should not be allowed to allows the nodes which are at a fair distance from other
drop below a threshold given by nodes to transmit at maximum power and also restrict the
power of nodes which are closer to each other. Now a node
with a packet to transmit is allowed to proceed with its
transmission if the transmission power will not disturb the
(2)
ongoing receptions in the node’s neighborhood beyond the
allowed interference margin. Allowing for concurrent
where,
transmissions increases the network throughput and
P t (i,j)= power transmitted by node i such that the
decreases contention delay.
transmission range does not exceed node j
P r (j,i)= power received by node j when node i transmits
P max = maximum power for this configuration

1) Spare Networks: The Locally Aware Protocol

This protocol is aimed at a solution to networks which are


sparse. Since the network is not crowded, one receiver does
not have multiple transmitters in its range. The
communication between a pair of nodes starts when the
transmitter senses the channel to be idle and sends a RTS to
the intended receiver at maximum power. This ensures that
the signal strength is high enough to maintain the SINR
level at the receiver incase of interference. This also helps
incase the intended receiver is farther away from the
transmitter. In such a case, increased power would mean
increased range. When the receiver hears the RTS
successfully, it waits for SIFS amount of time and responds
back with a CTS frame. The receiver calculates the ideal
transmitting power that the transmitter should use to carry
on further communication. This calculation is based on the Figure 2. A receiving node transmitting its collision
noise level at the receiver and the required SINR level to avoidance information in all the directions for the “Globally
process a signal correctly. This information is included in Aware” protocol
24 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 6, June 2010

It maintains these routes as long as they are needed by the


sources. Additionally, AODV forms trees which connect
3) Clusters: A Power Management Protocol multicast group members. The trees are composed of the
group members and the nodes needed to connect the
The cluster approach is defined for networks which tend to members. AODV uses sequence numbers to ensure the
operate in small groups. If all the nodes were to transmit at freshness of routes. It is loop-free, self-starting, and scales to
a random level of transmission power, the network large numbers of mobile nodes.
throughput would degrade since the neighboring nodes
would cause interference. It would be appropriate in such a AODV builds routes using a route request / route reply
scenario to restrict the power levels of different groups. query cycle. When a source node desires a route to a
Each member in a group uses the same power level. destination for which it does not already have a route, it
Different groups use different power levels depending on the broadcasts a route request (RREQ) packet across the
size of the group. Since new members may be added to the network. Nodes receiving this packet update their
cluster and old members might leave, a special node called information for the source node and set up backwards
the Cluster Head (CH) is elected. This node has knowledge pointers to the source node in the route tables. In addition to
of the topology of the network. It calculates the minimum the source node's IP address, current sequence number, and
required power so that all node pairs can communicate. This broadcast ID, the RREQ also contains the most recent
value is beaconed after short durations so that all the nodes sequence number for the destination of which the source
have the updated information and are transmitting with only node is aware. A node receiving the RREQ may send a route
the required power. Fig. 3 shows the reason for having each reply (RREP) if it is either the destination or if it has a route
group use its own power level. Nodes can still communicate to the destination with corresponding sequence number
within the cluster and use much less power. Since reducing greater than or equal to that contained in the RREQ. If this
the power also means a reduction in range for the is the case, it unicasts a RREP back to the source.
directional antennas, we see a better throughput result in our Otherwise, it rebroadcasts the RREQ. Nodes keep track of
case. This is a result of better spatial reuse. Next we study the RREQ's source IP address and broadcast ID. If they
the changes made to the Network layer. receive a RREQ which they have already processed, they
discard the RREQ and do not forward it. As the RREP
propagates back to the source, nodes set up forward pointers
to the destination. Once the source node receives the RREP,
it may begin to forward data packets to the destination. If
the source later receives a RREP containing a greater
sequence number or contains the same sequence number
with a smaller hop count, it may update its routing
information for that destination and begin using the better
route. As long as the route remains active, it will continue to
be maintained. A route is considered active as long as there
are data packets periodically travelling from the source to
Figure 3. Nodes inside a cluster can use lesser power to the destination along that path. Once the source stops
communicate sending data packets, the links will time out and eventually
be deleted from the intermediate node routing tables. If a
link break occurs while the route is active, the node
3.2 Power Control: The Network Perspective
upstream of the break propagates a route error (RERR)
Network layer solutions are energy-oriented and aim at message to the source node to inform it of the now
reducing energy consumption, with throughput being a unreachable destination(s). After receiving the RERR, if the
secondary factor. It includes intelligent routing based on source node still desires the route, it can reinitiate route
immediate link costs i.e. power aware routing [10]. Every discovery.
node in the network can compute the power consumed to With power control, AODV manages to preserve the
send a packet to its one hop neighbors. This allows nodes to network life as it routes packets through nodes which have
choose the path which consumes least power, thereby higher power remaining. This means when the AODV
preserving the system power. We implement the power protocol calculates the next hop on receiving a RREQ
aware AODV protocol and simulation results show that packet, it considers the remaining power of all its
network throughput increased marginally and system neighboring nodes before sending a RREP packet. In such a
lifetime was preserved. The advantage of power aware case every node must be periodically updated of the power
routing is that it can be integrated with the changed MAC levels of the neighboring nodes. We do so by periodically
layer, to enhance the system performance. The novelty here broadcasting the remaining power information to immediate
lies in the physical layer, with directional antennas. neighbors. If a node receives information that a neighboring
The Ad-hoc On Demand Distance Vector (AODV) routing node is running out of power, it must immediately calculate
algorithm is a routing protocol designed for ad-hoc mobile a new route to the destination. Although this protocol may
networks. AODV is capable of both unicast and multicast face slightly larger delays (since longer routes may be
routing. It is an on demand algorithm, meaning that it chosen depending on the remaining power of neighbors), it
builds routes between nodes only as desired by source nodes.
(IJCNS) International Journal of Computer and Network Security, 25
Vol. 2, No. 6, June 2010

manages to preserve network lifetime with the same 5. Results and Analysis
throughput efficiency.
We start this section by testing our protocol and studying the
4. Simulation Setup behavior of the network in different situations. We obtained
data by varying different parameters like the CBR rate and
Table 1 lists the parameters and their respective values that changing the number of nodes in a network. The CBR rate
have been used in the simulations. We simulate the three corresponds to how fast the packets are being generated and
topologies in NS2 on a Unix platform. The codes have been transmitted from the transmitter. Controlling this parameter
written in Tcl. In the next section we evaluate the means controlling the network load. Since a given network
performance of the cross layer design. We perform has only limited bandwidth, by increasing the CBR, the
simulations to obtain the network throughput and the throughput should first increase as more and more packets
average consumption of power in the network using our are sent successfully over the network. As the bandwidth is
design. The packet size is 512 bytes unless specified constrained in a network, on increasing the CBR rate
otherwise. Each flow in the network transmits CBR traffic. further, collisions occur, packets get dropped and
We do not consider mobility in our simulations. For the throughput decreases. So the network behaves best for a
radio propagation model, the two ray path loss model is particular CBR for a given scenario. This claim is validated
used. NS2 uses the 802.11 model for wireless networks. The by Fig. 4, 5 and 6. It is clear that the throughput in all the
power consumption has been calculated by considering the three types of networks increases as the CBR rate is
power used for transmission only, taking into consideration increased, but starts to decrease and attains a steady state
the number of directional antennas used. Perl language has when the rate is increased further. When the available
been used to calculate values from the trace files. The power bandwidth of a network is fixed and the number of nodes is
consumed for receiving and carrier sensing have not been varied, the affect is very similar to changing the CBR. With
calculated. For our simulations, we consider a mobile ad-hoc a constant CBR, when nodes are added to a particular
network in a square area of dimensions 500m X 500m. network, they increase the throughput of the network as
more packets are sent over the medium. It is noticed that
Table 1: Parameters and their values used in simulations when too many nodes start talking in the same space,
collisions occur and the throughput decreases. From Fig. 7,
TOPOLOGY PARAMETER VALUE 8 and 9 we can see that as we increase the number of nodes
in the three cases, the throughput increases to a maximum
Sparse Network Number of nodes 8 value. When a high number of nodes are present is any
Size of network 500 X 500 network, the value of throughput is much low. For Fig. 7, 8
Simulation time 500 secs and 9 the number of nodes in each network is specified in
Packet size 512 bytes Table I. The CBR rate for the sparse network is 20 Kbps and
Bandwidth 1Mb for the dense and cluster network is 40 Kbps. These tests
Traffic Type CBR show that even after making changes to the MAC, Network
Pmax 100mW and Physical Layer of the network, it has good behavior.
MinRecvPwr 1mW
Dense Network Number of nodes 30 It should be noted that in Fig. 7, when there are 18 nodes in
Size of network 500 X 500 the network, the throughput is very low. This is attributed to
Simulation time 500 secs the fact that the cross-layer design for the sparse network is
not effective when a large number of nodes are present. This
Packet size 512 bytes
is the point where we switch from the Locally Aware
Bandwidth 2Mb
Scheme to the Globally Aware Scheme and we say our
Traffic Type CBR
network is no longer sparse. Using the Globally Aware
Pmax 100mW Scheme helps more nodes in the network be aware of the
MinRecvPwr 1mW SINR values at the receivers and they can then adjust their
Cluster form Number of nodes 16 transmission power accordingly. Different networks may
Size of network 500 X 500 have a different value for number of nodes in sparse and
Simulation time 500 secs dense network.
Packet size 512 bytes
Bandwidth 2Mb
Traffic Type CBR
Pmax 100mW
MinRecvPwr 1mW
26 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 6, June 2010

Figure 4. Sparse Network: Throughput values for different


CBR rate. Figure 7. Sparse Network: Throughput values with different
number of nodes.

Figure 5. Dense Network: Throughput values for different


CBR rate.
Figure 8. Dense Network: Throughput values with different
number of nodes.

Figure 6. Cluster Network: Throughput values for different


CBR rate.

Figure 9. Cluster Network: Throughput values with


different number of nodes.
(IJCNS) International Journal of Computer and Network Security, 27
Vol. 2, No. 6, June 2010

Since we are aiming at conserving power and increasing the


throughput, it is necessary to compare the performance of
these 3 types of topologies when they are not empowered by
the cross layer design. In traditional networks, omni
directional antennas are used. We are going to make the
following comparisons: 1) between a network equipped with
omni directional antennas (802.11b) to a network using
directional antennas with CLPM implementation, 2)
between a network equipped with directional antennas
(802.11b-like) to a network using directional antennas with
CLPM implementation. The terms SN, DN and CN imply
Sparse Network, Dense Network and Cluster Network
respectively. In the cases where there is no power control,
all the nodes transmit at the maximum allowable
transmitting power. Simulation results for five different
networks in each topology case (namely sparse, dense and Figure 11. Throughput comparison between the three
cluster) have also been shown at the end (Fig. 14, 15 and topologies when using a regular directional antenna and a
16) to give the reader an idea of the variation in results. directional antenna with our power scheme implemented.
Fig. 10 compares the throughput between the three In Fig. 12 we have compared the power consumed between
topologies. It is to be kept in mind that the available the three topologies between a network having omni
bandwidth for the sparse network was 1Mb while it was directional antennas and a network having directional
2Mb for the dense network and the cluster form. The cluster antennas with power control. It is clear from the results that
form network shows maximum improvement in throughput in case of all the three topologies, the network with a power
(13.33% more) which indicates that when all nodes in the control scheme performs better than a network with omni
cluster are allowed to transmit without power control, it directional antennas. The maximum allowable transmission
leads to many collisions. power for a node in any case was 100mW and the
MinRecvPwr was 1mW.

Figure 10. Throughput comparison between the three Figure 12. Comparison in power consumed in different
topologies when using a omni directional antenna and a topologies when using a omni directional antenna to using a
directional antenna with our power scheme implemented. directional antenna with power control.

Fig. 11 shows that implementing the power control design Fig. 13 shows the difference in power consumed in case of
increases the throughput by nearly 6% (in the cluster form the three topologies when using a regular directional
case) when compared to a directional antenna without power antenna to when a directional antenna with power control is
control. This is attributed to the fact that reducing the used. Controlling the transmission power shows a lower
transmission power allowed for more node pairs to value of consumed power as compared to a network when
communicate and thus more packets were transmitted all the nodes are transmitting at the maximum power. The
successfully. The dense network and the cluster form simulations show that reducing the transmission power of
topologies also show better performance once the cross layer the nodes did not affect the throughput of the system
design in used. negatively and also helped in conserving the overall network
power. Throughout the simulation, the receiver keeps the
transmitter aware of the ideal transmitting power by sending
its SINR value to the transmitter. The network without
28 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 6, June 2010

power control uses nodes that transmit with maximum


power all the time.

Figure 16. Simulation results for power consumption for


five different network scenarios used in the Cluster form
topology case.
Figure 13. Comparison in power consumed in different 6. Conclusion
topologies when using a regular directional antenna to using
a directional antenna with power control. We proposed a CLPM (cross layer design for power
management) and throughput enhancement for wireless ad-
This leads to wastage of energy and also hampers hoc networks. Our primary objective was to exploit the
conversations going on in the nearby space. The overall advantages of directional antennas and develop a power
network power is saved by almost 50% in the sparse control scheme for a sparse, a dense and a clustered
topology and also by considerable amounts in the dense and network. The underlying MAC layer is responsible for
the cluster form. controlling the transmission power of a node depending
upon the minimum required SINR level at the receiver. This
information is sent by the receiver to only the intended
transmitter incase of sparse networks. In case of dense
networks, this collision avoidance information is transmitted
in all the directions in lieu of more transmitters in a
receivers range. This helped in not only conserving the
power of nodes but also increased the throughput as more
communications could take place in the same space. The
Network layer performs a power aware routing protocol
called Power AODV which aims at reducing the overall
power consumed in routing packets. Performance
comparison of this cross layer design with a 802.11b
network and a 802.11b-like network using directional
antennas, showed that it reduced the overall network power
consumption and also increased the throughput. Thus, a
network with a power management scheme implemented
Figure 14. Simulation results for power consumption for will have better performance than a network without such a
five different network scenarios used in the Sparse topology scheme.
case.
References

[1] Rong Zheng, Robin Kravets, “On-demand Power


Management for Ad-hoc Networks”, INFOCOM
2003.
[2] Zongpeng Li and Baochun Li, “Probabilistic Power
Management for Wireless Ad-hoc Networks”, Mobile
Networks and Applications, October 2005.
[3] Krishnamurthy, Srikanth ElBatt, Tamer Connors,
Dennis, “Power management for throughput
enhancement in wireless ad-hoc networks”, In
Proceedings of the IEEE ICC 2000.
[4] Benjie Chen, Kyle Jamieson, Hari Balakrishnan, and
Robert Morris, “Span: An Energy-Efficient
Figure 15. Simulation results for power consumption for Coordination Algorithm for Topology Maintenance
five different network scenarios used in the Dense topology in Ad-hoc Wireless Networks”, Wireless Networks
case. archive Vol. 8, Issue 5, September 2002.
(IJCNS) International Journal of Computer and Network Security, 29
Vol. 2, No. 6, June 2010

[5] Vikas Kawadia and P. R. Kumar, “Power Control and Authors Profile
Clustering in Ad-hoc Networks”, INFOCOM 2003.
[6] El-Osery, A.I. Baird, D. Bruder, S, “Transmission Mahasweta Sarkar joined the department
power management in ad-hoc networks: issues and of Electrical and Computer Engineering at
advantages”, In Proceedings of the IEEE Networking, San Diego State University as an Assistant
Sensing and Control, 2005. Professor in August 2006, after receiving
[7] Marwan Krunz and Alaa Muqattash, “Transmission her Ph.D. from UCSD in December 2005.
She received her Bachelor’s degree in
Power Control in Wireless Ad-hoc Networks:
Computer Science (Suma Cum Laude)
Challenges, Solutions, and Open Issues”, Network, from San Diego State University, San
IEEE Sept-Oct 2004. Diego in June, 2000. Prior to joining
[8] Matthias Rosenschon1, Markus Heurung1, Joachim SDSU, she worked as a Research Scientist
Habermann, “New Implementations into Simulation at SPAWAR Systems Center, Point Loma, San Diego. Her
Software NS-2 for Routing in Wireless Ad-Hoc research interests include protocol designs, flow control,
Networks”, 2004. scheduling and power management issues in Wireless Local Area
[9] http://www.isi.edu/nsnam/ns/tutorial/ Networks.
[10] Asis Nasipuri, Kai Li and Uma Reddy Sappidi, “Power
Consumption and Throughput in Mobile Ad-hoc Ankur Gupta is currently a graduate
research student under Dr. Mahasweta
Networks using Directional Antennas”, In
Sarkar at San Diego State University, San
Preceedings of the Eleventh International Diego. He received his Bachelor’s degree in
Conference, Computer Communications and Electronics Engineering in June 2001 from
Networks, 2002. Pune University, India. His research
[11] Thanasis Korakis, Gentian Jakllari and Leandros interest includes analysing and studying the
Tassiulas, “A MAC protocol for full explotiation of effect of power management protocols in
Directional Antennas in Ad-hoc Wireless Networks”, wireless devices that emply directional
MobiHoc, June 1-3, 2003, Annapolis, Maryland, antennas.
USA.
[12] ANSI/IEEE Std 802.11, “Wireless LAN Medium
Access Control (MAC) and Physical Layer (PHY)
specifications”, 1999.
[13] Z. Huang, Z. Zhang, and B. Ryu, "Power control for
directional antenna-based mobile ad-hoc networks",
in Proc., Intl. Conf on Wireless Comm. and Mobile
Computing, pp. 917-922, 2006.
[14] Ram Ramanathan, "On the performance of ad-hoc
networks with beamforming antennas", Proc. Of
MobiHoc '200 1, pages 95-105, October 2001.
[15] B. Alawieh, C. Assi, W. Ajib, "A Power Control
Scheme for Directional MAC Protocols in MANET",
IEEE Conf, WCNC2007, pp.258-263, 2007.
[16] Nader S. Fahmy, Terence D. Todd, and Vytas Kezys,
"Ad-hoc Networks with Smart Antennas Using IEEE
802.11-Based Protocols", In Proceedings of IEEE
International Conference on Communications (ICC),
April 2002.
[17] P. Karn., “MACA -A New Channel Access Method for
Packet Radio”, in Proc. 9th ARRL Computer
Networking Conference, 1990.
[18] E.-S. Jung and N. Vaidya, “A power control MAC
protocol for ad-hoc networks”, In 8th ACM Int. Conf
on Mobile Çomputing and Networking
(MobiCom02), pages 36-47, Atlanta, GA, September
2002.
[19] J.-P. Ebert, B. Stremmel, E. Wiederhold, and A.
Wolisz, “An Energy-efficient Power Control
Approach for WLANs”, Journal of Communications
and Networks (JCN), 2(3): 197-206, September 2000.
[20] Mineo Takai, Jay Martin, Aifeng Ren, and Rajive
Bagrodia, "Directional virtua1 carrier sensing for
directiona1 antennas in mobile ad-hoc networks",
MOBIHOC'02, June 2002.

S-ar putea să vă placă și