. RESEARCH PAPER .

SCIENCE CHINAInformation Sciences

June 2012 Vol. 55 No. 6: 1360–1371

doi: 10.1007/s11432-012-4550-6

c© Science China Press and Springer-Verlag Berlin Heidelberg 2012 info.scichina.com www.springerlink.com

An adaptive directional MAC protocol for ad hocnetworks using directional antennas

LU XiaoFeng1∗, TOWSLEY Don2, LIO Pietro3 & XIONG Zhang4

1School of Computer Science, Beijing University of Posts and Telecommunications, Beijing 100876, China;2Deptartment of Computer Science, University of Massachusetts, Amherst 01003, USA;3Computer Laboratory, University of Cambridge, Cambridge CB3 0FD, United Kingdom;

4School of Computer Science, Beihang University, Beijing 100191, China

Received February 4, 2011; accepted August 18, 2011; published online March 12, 2012

Abstract A directional antenna can bring benefits in terms of power consumption, spatial reuse, etc. To

exploit the advantage of directional antennas and improve the transmission throughput highest, this paper

proposes an adaptive directional MAC protocol (ADMAC). By varying the transmission strategy according

to the usage of the channel, nodes can send RTS and CTS packets omni-directionally or directionally. Also,

this paper proposes a calculation method of virtual carrier sensing with collision avoidance. By the method,

ADMAC protocol makes more pairs of nodes transmit and receive data simultaneously without interferences

than other MAC protocols for directional antennas. The paper compares the simultaneous delivering nodes and

network throughput of ADMAC with DMAC, DVCS and SDMAC under different experiment parameters. The

simulation results show that the throughput of ADMAC is higher than the throughput of DMAC, DVCS and

SDMAC protocols.

Keywords medium access control, directional MAC, directional antennas, MANETs

Citation Lu X F, Towsley D, Lio P, et al. An adaptive directional MAC protocol for ad hoc networks using

directional antennas. Sci China Inf Sci, 2012, 55: 1360–1371, doi: 10.1007/s11432-012-4550-6

1 Introduction

A mobile ad hoc network (MANET) is a wireless network without fixed base station or any wireline back-

bone infrastructure. Nodes use peer-to-peer packets transmission and multi-hop routes to communicate.

MANETs are used in collaborative, distributed mobile computing and especially in scenarios where wired

networks are simply ineffective or implausible, such as disaster recovery and survival search.

Typically, a common assumption on ad hoc networks is that nodes are equipped with omni-directional

antennas. An omni-directional antenna has a 360◦ coverage angle. It sends signals towards all horizontal

directions [1]. However, since the energy is broadcasted in all directions and only a very small portion

is actually received by the intended node, most of the energy transmitted is wasted [2,3]. Directional

antennas have several advantages over omni-directional antennas. A directional antenna can spread the

energy to a certain direction, so it decreases the energy consumed by the transmitter [4–6]. A directional

antenna can increase the potential spatial reuse and network capacity [7–9]. Directional antennas can

∗Corresponding author (email: luxf@bupt.edu.cn)

Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6 1361

C

A B D C

A B

C

A B

(a) (b) (c)

Figure 1 An illustration of the hidden terminal problem caused by directional antennas.

bring benefits in the security of routing by reducing the probability of being detected by detection systems

[10]. Some researchers applied the directional antennas in VENET to improve the performance of the

communication between the vehicles and the roadside APs [11].

As the wireless channel is shared by all users, a medium access control (MAC) protocol is needed to

coordinate different users to share the wireless channel impartially and efficiently. The IEEE 802.11 DCF

(distributed coordination function) is a widely used MAC protocol for wireless ad hoc communications

[12–14]. IEEE 802.11 was designed for omni-directional antennas, but its carrier sense multiple acess

with collision avoidance (CSMA/CA) is the foundation of many directional MAC protocols.

The use of directional antennas causes some new problems, such as finding the direction of the receiver,

the new hidden terminal problem and the head-of-line blocking problem [15–18].

1.1 Finding the direction of the receiver

When a node employs a directional antenna to transmit data, the node must point its direction antenna

at the receiver. If nodes are mobile, the transmitter has to vary the direction of its directional antenna

frequently according to the receiver’s movement. When nodes are moving at high speeds, finding the

direction of the receiver all the time is difficult.

1.2 New hidden terminal problem

Hidden terminal problems have been studied for IEEE 802.11 MAC protocol using omni-directional

antennas. In Figure 1(a), node A and B employ omni-directional antennas to send data. Node C can

receive the Request To Send (RTS) packet from node A and the Clear To Send (CTS) packet from

node B, so it will not send RTS to node A or B when they are communicating. However, when they

use directional antennas to communicate, the hidden terminal problem becomes serious. In Figure 1(b),

node A sends a directional-RTS (DRTS) to node B, and node C does not receive the DRTS. Also, node

C does not receive the CTS packet sent by node B to A because it is beamforming in a different direction

to node D. During the time when A is sending data to B, if C finishes the transmission with D and has

data for B, the transmission can be allowed because node C does not know that A is communicating with

B. The transmission will interfere with the ongoing transmission from A to B as Figure 1(c) shows. This

is the new hidden terminal problem caused by directional antennas.

1.3 Head-of-line blocking problem

If the channel in the outgoing direction of the head of the RTS queue is in use, the node will have to

wait until the channel is available to send the head RTS packet. In fact, during the blocking time, the

channels in the outgoing directions of other RTS packets in the RTS queue are idle probably. Thus, the

head of RTS queue blocks the transmission of other packets, which is the head-of-line blocking problem.

Head-of-line blocking problem may make the throughput of some directional MAC protocols as low as

IEEE 802.11 in some scenarios.

The objective of the proposed MAC protocol is to make the channel be used more efficiently and to

improve the network throughput. By the directional MAC protocol we proposed, a transmitter selects the

1362 Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6

RTS packet in whose outgoing direction the channel is idle. Each node in the proposed MAC protocol has

a neighbor table to save all its neighbors’ directions. So when a node wants to send data to a neighbor,

it can get the neighbor’s direction and point its directional antenna at this neighbor directly. When a

node receives a packet whose target is another node, it updates this neighbor’s current direction noted in

the neighbor table. Also, we address the collision avoidance calculation based on the directional network

allocation vector (DNAV) and the neighbor table.

The contributions of this paper are that we propose an adaptive directional MAC (ADMAC) protocol

for directional antennas to improve the channel utilization rate. The adaptive directional MAC protocol

includes four parts: (1) directional virtual carrier sensing with collision avoidance, (2) adaptive access

channel mechanism, (3) adaptive neighbor discovery, (4) neighbor table update. Nodes can select their

transmission strategies to send the RTS and CTS packets omni-directionally or directionally and deter-

mine the direction in which the control packet is to be sent. The experiment shows that ADMAC can

improve the wireless channel utilization rate and the network throughput significantly.

This paper is organized as follows. We review some related work on directional transmission protocols

in Section 2 and introduce the antenna model we employ in Section 3. We introduce the proposed

directional MAC protocol in Section 4. We evaluate the network transmission performance of ADMAC

in Section 5 and conclude our work in Section 6.

2 Related work

In the past, there were several studies regarding the MAC protocols for MANETs using directional

antennas.

Nasipuri et al. proposed a simple MAC protocol for nodes that were equipped with multiple directional

antennas [19]. In their study, they assumed that a node was equipped with M directional antennas. The

transmitter sends RTS packets by M directional antennas towards all directions, so the receiver’s M

directional antennas can receive several transmissions from different directions. The receiver sends omni-

directional CTS as well. Then the transmitter sends data and the receiver receives data directionally.

This protocol does not fully utilize the advantage of directional antennas because the spatial use is almost

the same as IEEE 802.11.

Ko et al. proposed a DMAC protocol for MANETs using directional antennas which includes two

schemes [20]. In scheme 1, the CTS packets are always transmitted omni-directionally, while the RTS

control packets are transmitted directionally. In scheme 2, both the CTS and RTS are transmitted

omni-directionally. The shortcoming of Ko’s DMAC is that the CTS packets are always transmitted

omni-directionally, and this will make the receiver have to wait for a long time. In the proposed ADMAC

protocol, the RTS is transmitted directionally but the CTS is transmitted either omni-directionally or

directionally according to the channel utilizing situation. The receiver can send multi-directional CTS

packets instead of sending an omni-directional CTS to avoid interfering others’ transmissions.

Choudhury et al.’s research focused on using multi-hop RTSs to establish links between distant nodes

and transmit data over a single hop [21]. However, it would cost some time to forward the receiver’s

location to the transmitter through many relays. In mobile networks, if nodes are mobile, when the

transmitter gets the receiver’s location after multi-hops, the receiver may move to other places. Therefore,

their MAC protocol is more adaptable for static ad hoc networks.

Minero Takai et al. proposed a directional virtual carrier sensing scheme for directional antennas, DVCS

[22]. By DVCS, the transmitter sends a RTS packet directionally to the receiver, and the receiver sends

a CTS packet to the transmitter directionally as well. DVCS protocol assumes that directional antennas

can know the direction of the arrival signals. This assumption is available for current antenna technology.

We employ the same assumption of directional antennas as DVCS. The objective of DVCS is to decrease

the reserved region. However, by DVCS protocol, as the CTS packet is transmitted directionally to the

transmitter, not all neighbors of the receiver can receive the CTS packet and know the receiver is in

communication, and some of them maybe want to send data to the receiver, which may interfere with

the receiver’s transmission.

Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6 1363

Bao et al. proposed a distributed receiver-oriented channel access scheduling protocol (ROMA) for

directional antennas [23]. ROMA protocol determines the links activation in every time slot using two-

hop topology information. A node updates its one-hop neighbors’ locations so that the next round of

channel access scheduling is free of errors. However, because nodes update their neighbors’ locations

omni-directionally, they may interfere with other’s transmission.

Recently, Li et al. have proposed a MAC protocol called SDMAC that improves the network throughput

[24]. There are two types of RTS and CTS packets in SDMAC: Type I and Type II. When a sender wants

to send data to a receiver, it sends a Type I directional-RTS (DRTS) to the receiver. The receiver will

send back a Type I directional-CTS (DCTS) packet to the sender. Now, both the sender and the receiver

know the new relative direction of each other, the sender sends Type II DRTS and the receiver sends

type II DCTS. Type II DRTS and DCTS cover all the neighborhoods of the sender and the receiver, and

can be sent out simultaneously without overlapping transmission. The network throughput of SDMAC

is better than existing MAC protocol using directional antennas. Using ADMAC protocol, nodes do

not need to send RTS and CTS two rounds, which can save time and improve the network throughout

further.

3 Antenna model

In our antenna model, we assume that an antenna can work in two modes: omni-directional mode and

directional mode. Antennas are either omni-directional mode or directional mode [1]. Omni-directional

antennas cover 360◦ and form an approximate circular radiation region. All nodes in the radiation region

can receive communication signals [2]. A directional antenna can form a directional beam pointing to

one direction by concentrating its transmit power into that direction. By pointing the main directional

beam at the receiver, a directional antenna provides additional gain in the direction of the receiver. We

assume the directional antenna can know the direction in which the signal strength is the strongest and

can steer its antenna towards that direction.

Antennas can send and receive data in both these two modes. If nodes have nothing to transmit,

their antennas work in the omni-directional mode to detect signals. Since directional antennas can get

additional gain in the main lobe direction, they consume less power to transmit data directionally than to

transmit data omni-directionally. A transmitter and receiver can communicate over a large distance when

both of them are in directional mode, as compared to when they are in directional and omni-directional

mode, respectively.

There are three primary types of directional antenna systems: switched beam antenna system, steered

beam antenna system, and adaptive antenna system [25]. In this study, we use the steered beam antenna

system. In fact, the proposed MAC protocol is also suitable for the other two directional antenna

systems. When the angle between two main lobe directions is wider than the beamwidth angle, the two

transmissions will not interfere with each other.

It is possible to make adaptive beam using arrays of radiating elements [26]. An array consists of

two or more antenna elements that are spatially arranged and electrically interconnected to produce a

directional radiation pattern. In our study, we assume the beamwidth of directional antennas is fixed,

but we will study the network throughput under different antenna beamwidths.

4 ADMAC: adaptive directional MAC protocol

The objective of the proposed adaptive directional MAC protocol is to improve network throughput

by making as many pairs of nodes as possible to communicate simultaneously. The proposed ADMAC

consists of (1) directional virtual carrier sensing with collision avoidance, (2) adaptive directional medium

access control, (3) adaptive neighbor discovery, (4) neighbor table update.

4.1 Virtual carrier sensing with collision avoidance

When a node receives an RTS packet from other nodes, it can know the ID and the direction of the

sender. The RTS packet does not mean the connection from the sender to the specific receiver has been

1364 Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6

A B

D

C

α

β

Figure 2 Angle α and β. α = ∠BCD, β = ∠ABC.

A

C

B

D

A

C

D

AC

B

D

C

A

D

B

(a) (b) (c) (d)

B

Figure 3 Collision avoidance.

established successfully. The node notes the sender’s ID and direction into its neighbor table. When a

node receives a CTS packet that can be an OCTS or DCTS, it knows that the specific receiver is prepared

to receive data from the sender and the transmission will start soon. Nodes employ DNAV to store the

reserved transmissions between other nodes. DNAV notes the receiver’s ID, the direction in which it

receives the CTS, the sender’s ID and the reserved transmission duration. If the node can know the

direction to the sender from its neighbor table, it writes the sender’s direction into the DNAV as well.

Otherwise the sender’s direction in the DNAV is empty.

Assume that a node wants to send data to another node while one of its neighbors is communicating.

Take the scenario shown in Figure 2 as an example where node B is sending data to node A. Now, node

C wants to send data to node D and it knows the direction of node A, B and D. Node C calculates two

angles to check whether its transmission will interfere with the transmitting from node B to node A. The

first angle is between node D’s direction and node B’s direction, which is referred to as α, α = ∠BCD.

The second is the angle between node B’s direction and the direction of the ongoing transmitting, which

is referred to as β, β = ∠ABC.

Assume the beamwidth of the directional antenna is ω. Node C checks whether its transmission will

interfere the ongoing transmiting from node B to node A by the following algorithm:

(1) α > ω/2, β > ω/2. It indicates node B is not in node C’s transmission range and node C is not

in node B’s transmission range neither, which is as showed in Figure 3(a). Node C can send the RTS to

node D.

(2) α < ω/2, β > ω/2. It indicates node B is in node C’s transmission range but node C is not in

node B’s transmission range, which is as showed in Figure 3(b). The transmission from node C will not

interfere with node B, so node C can send the RTS to node D.

(3) α > ω/2, β < ω/2. It indicates node B is not in node C’s transmission range but node C is in

node B’s transmission range, which is as showed in Figure 3(c). The transmission from node C will not

interfere with node B.

(4) α < ω/2, β < ω/2. It indicates node B is in node C’s transmission range and node C is in node B’s

transmission range either, which is as showed in Figure 3(d). The transmission from node C will interfere

with node B, so node C cannot send the RTS to node D.

Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6 1365

4.2 Adaptive medium access control

4.2.1 Send a DRTS adaptively

When a node needs to send a RTS packet, it selects the head of its RTS queue and checks whether

the channel in the outgoing direction is idle. If the channel is idle, the sender will check whether the

transmission will interfere with other ongoing transmissions. If its transmission does not interfere with

any ongoing transmission, the sender sends a directional RTS (DRTS) in its requested direction. If the

channel is in use, the sender will select the next of the RTS queue and check its outgoing channel again.

In a word, the sender selects the first unblocking RTS packet in the queue to send. This can resolve the

HOL blocking problem. The algorithm for a sender to send RTS packet is as follows. The RTS packet

includes the sender’s ID, the specific receiver’s ID and the expected duration of the transmission.

Begin

p=RTSQueue.head;

while (p!=null) do

if (Channel(p)==idle) then

if (CheckCollision(p,DNAV)==0) then

Send(p);

end if

p=p.next;

else

p=p.next;

end if

if (p=null) then

p=RTSQueue.head;

Wait;

end if

end while

End

4.2.2 DRTS and OCTS shake-hands

When the specific neighbor receives the DRTS packet, it sends back a CTS packet to the sender. A CTS

packet contains the proposed duration of data transmission and the sender and receiver’s IDs as well. If

the receiver’s 360◦ channels are all free, it sends an omni-CTS (OCTS) packet and all the neighbors of

the receiver can receive the OCTS packet. They write the transmitter’s ID, receiver’s ID, the reserved

transmission duration and the direction of the receiver into their DNAVs. These neighbors will not send

packets to the receiver or sender during the reserved duration. Thus the receiver’s communication will

not be interfered.

After the receiver sends an OCTS packet, it changes its antenna mode into the directional mode and

points the directional antenna at the transmitter to receive data. The sender sends data to the receiver

directionally. Receiver sends an ACK packet to the transmitter every N data packets. The ACK packet

indicates the data sequence number that the receiver has received correctly.

In Figure 4, assume that node B wants to send data to node C. Firstly, node B sends a DRTS to node

C. After node C receives the DRTS, it sends an OCTS omni-directionally and all its neighbors including

node B can receive the OCTS. When node D receives the OCTS from node C, it knows that node C will

receive data from node B and the reserved transmission duration. Thus, node D will not send DRTS

packet to node C during the reserved transmission duration.

4.2.3 DRTS and DCTS shake-hands

Let us consider a scenario where when a node receives a DRTS and wants to send a CTS to the sender,

the channels in one direction or some directions are in use, so the receiver cannot send the CTS omni-

directionally. However, the receiver will not hold the expected transmission of CTS just depending on

1366 Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6

E

A B

OCTS OCTS

DRTS

DC

OCTS

OCTS

DA B

C

Figure 4 An illustration of DRTS and OCTS shake-

hands. Node B sends a DRTS to node C, and node C

sends an OCTS in all directions. Node D will not send an

RTS to node C during the reserved transmission duration.

Figure 5 An illustration of DRTS and DCTS shake-

hands: a directional RTS and a directional CTS.

DCTS

AD

OCTS

DRTS

CB

Figure 6 An example of a rightabout DCTS: The transmitter, node B, sends a DCTS to node A which is in the reverse

direction from B to C to avoid the potential collision.

the DNAV, it checks whether the expected transmisssion will interfere with the ongoing transmission

according to the method introduced in Section 4.1. For example, in Figure 5, node A and B are in

communication and node C receives a DRTS from node D during the transmiting between A and B.

Before node C sends the CTS packet, it checks its DNAV and finds that its neighbor node B is receiving

data from A. Then, node C calculates the angle between−−→BC and

−−→CD and

−−→AB, say the angle ∠BCD and

∠ABC. Because ∠BCD > ω/2 and ∠ABC > ω/2, node C sends a DCTS to node D.

4.2.4 The sender sends a DCTS rightabout

New hidden terminal problem can be caused by directional transmissions. In Figure 6, node B sends

DRTS to node C and node A can neither receive the DRTS packet from node B nor receive the CTS

packet from node C because it is out of node C’s transmission range. So node A does not know that node

B is sending data to node C. If node A wants to communicate with node B, it sends a DRTS to node A.

Although node C is out of node A’s transmission range, node A’s transmission still interferes with node

C’s transmission because the interference range is larger than the transmission range.

With the proposed ADMAC protocol, if node B finds that there are nodes in the rightabout direction

from itself to the receiver, the transmitter, say node B, sends a DCTS in the reverse direction from itself

to the receiver to avoid the potential hidden terminal problem caused by node A. As node A receives the

DCTS from node B, A will not send the RTS packet towards node B. If node A wants to send data to

node D which is out of node B’s transmission range as Figure 6 shows, the transmission between A and

D will not interfere with the transmission between B and C.

4.3 Adaptive neighbor discovery

When a node wishes to send data to another node which is not in its neighbor table, it starts a neighbor

discovery process to find this node’s direction. The neighbor discovery of directional antennas is as

follows.

Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6 1367

ORTS

T

ORTS

CTS

R Transmission

DRTS

DRTS

(a) (b)

Transimission

ORTS

T

ORTS

CTS

R Transmission

DRTS

DRTS

(a) (b)

Transimission

T

Figure 7 An illustration of neighbor discovery. (a) Omni-directional neighbor discovery; (b) directional neighbor discovery.

(1) If there is no ongoing transmission within its neighborhood, the transmitter broadcasts a RTS

packet omni-directionally as Figure 7(a) shows. If there are ongoing transmissions within the transmitter’s

neighborhood as Figure 7(b) shows, the transmitter cannot broadcast an ORTS packet. It sends several

DRTS packets towards the directions in which there are no transmissions. After the transmitter sends a

RTS packet in one direction, it waits for the CTS packet from the desired receiver for SIFS.

(2) When the desired receiver receives the ORTS or DRTS packet, it will send an OCTS packet if the

channel is free, or it will send several DCTS to directions where there is no transmission.

(3) When the transmitter receives CTS packet from the receiver, it knows the receiver’s direction. It

fixes its directional antenna towards the receiver and prepares to transmit data. If it does not receive the

CTS packet within SIFS duration, it rotates its antenna a beamwidth angle and repeats step (1).

(4) If the transmitter does not receive any CTS packet from the receiver after its directional antenna

has rotated a round, it waits for the duration of steering a round and repeats above steps until the repeat

times are larger than a threshold, such as 16.

Besides, during the transmission between two nodes, a node may move out of the other node’s trans-

mission region, and then it cannot receive data any more. When the receiver finds that it cannot receive

data from the transmitter, it will send a re-send packet to the transmitter at the reserved ACK time and

indicate the sequence number that the transmitter should re-send. If the transmitter receives the re-send

packet, it modifies the direction of its directional antenna and resends data. If the transmitter neither

receives the supposed ACK packet nor receives a re-send packet, it has to discover the receiver’s direction

again. However, the receiver cannot move far away from its last location during a short time. Therefore,

the new direction in which the sender sends RTS packet should not be far away from the last direction in

which the transmitter received ACK packet successfully. The transmitter rotates its directional antenna

left and right a beamwidth angle from the last direction and sends DRTS packets again as shown in

Figure 8. After the transmitter finds the receiver, it re-sends the data which was not received correctly.

4.4 Neighbor table update

By ADMAC protocol, each node has a neighbor table to store its neighbors’ directions. The neighbor

table records each neighbor’s ID, last direction, current location. When a node receives a packet from one

of its neighbors, it can know the ID and direction of this neighbor. Then, the node writes the direction

into the neighbor table as this neighbor’s current direction.

If a node receives a RTS or CTS packet from a neighbor that is already in the neighbor table, it transfers

the content in the current direction field to the last direction field and inputs the new direction into the

current direction field. Thus, before a node wants to send data to its neighbor, it checks its neighbor

table to find this neighbor’s direction. Moreover, the node can compare the directions in the last direction

field and current direction field to guess this neighbors’ moving direction. It can help the transmitter to

determine the direction of the receiver and make the DRTS be received with high probability.

1368 Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6

T

Last direction

Turn left

DRTS

RDRTS

Beam width

Turn right

Figure 8 Neighbor recovery. The transmitter rotates its antenna left/right a beamwidth angle and sends DRTS packets.

(a) (b)

DMAC DVCS SDMAC ADMACDMAC DVCS SDMAC ADMAC

120

100

80

60

40

20

0

Sim

ulta

neou

s de

liver

ing

node

s

120

100

80

60

40

20

0

Sim

ulta

neou

s de

liver

ing

node

s

50 100 150 200 250 50 100 150 200 250

The number of nodes, R=100 The number of nodes, R=200

Figure 9 The simultaneous delivering nodes under different nodes densities. (a) R=100 m; (b) R=200 m.

5 MAC performance evaluation

5.1 Metrics and simulation

Directional communication makes the transmitter send data and the receiver receive data directionally.

Besides, it ensures that different transmissions do not interfere with each other, so different transmissions

can go on simultaneously. Here, we employ simultaneous delivering nodes (SDN) and Throughput to

measure the transmission performance of MAC protocols. The number of simultaneous delivering nodes

is the number of nodes that can communicate simultaneously.

The simulation area is a 2000 m × 2000 m square area and nodes are located in the area randomly.

Since a directional antenna gets additional antenna gain in the main lobe direction, the transmission

range and beamwidth of the main lobe would affect the throughput. The beamwidth of a directional

antenna is 20◦. We assume the transmission ranges of all nodes to be the same, but the transmission

range is not fixed. In the simulation, we randomly select nodes as the transmitters and try to make

more pairs of nodes be able to communicate simultaneously. However, how many nodes of the nodes can

send data is determined by the MAC protocols. We compare the network throughput of ADMAC with

DMAC, DVCS and SDMAC.

5.2 Simulation results

Figure 9 shows the number of simultaneous delivering nodes under different number of nodes of four

MAC protocols. From this figure, what we can know is that with the increasing of nodes involved

in the simulation, the numbers of simultaneous delivering nodes of the four MAC protocol increase.

Under a given transmission range (R), the simultaneous delivering nodes of ADMAC are more than

those of DMAC, DVCS and SDMAC. Also, we find that under the given number of nodes, the number of

Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6 1369

ADMACDMAC DVCS SDMAC ADMAC

(a) (b)

DMAC DVCS SDMAC

25

20

15

10

5

0100 200 300 400 500100 200 300 400 500

Transmission range Transmission range

140

120

100

80

60

40

20

0Sim

ulta

neou

s de

liver

ing

node

s

Sim

ulta

neou

s de

liver

ing

node

s

Figure 10 The simultaneous delivering nodes under different transmission ranges. (a) N=50; (b) N=250.

simultaneous delivering nodes by DMAC and DVCS both decreases with the increasing transmission

range, say from 100 to 200 m. In fact, with the increasing transmission range, more nodes can receive

other nodes CTS and RTS packets, so more nodes have to wait for the reserved transmission duration by

DMAC and DVCS. However, the number of simultaneous delivering nodes by SDMAC and ADMAC when

R=200 m is more than that when R=100 m under given number of total nodes. When the transmission

range is 100 m and the total number of nodes is 250, the number of simultaneous delivering nodes of

DMAC is 64 and that of DVCS is 68, the number of simultaneous delivering nodes of SDMAC and

ADMAC are 85 and 89. When the transmission range is 200 m and the total number of nodes is 250,

the number of simultaneous delivering nodes of DMAC is 33 and that of DVCS is 42, the number of

simultaneous delivering nodes of SDMAC and ADMAC are 97 and 118, which are remarkably larger than

that of DMAC and DVCS.

Figure 10 shows the number of simultaneous delivering nodes under different transmission ranges (R).

SDN of ADMAC is always larger than that of the other three MAC protocols under any transmission

range. A valuable conclusion is that given the number of nodes, SDN can reach the maximum value at a

specific trnasmission range. In Figure 10(a), SDN of DMAC reachs the maximum value when R = 100,

SDN of DVCS is largest when R = 300, SDN of SDMAC is largest when R = 400 and ADMAC gets the

largest SDN when R = 500. However, when the nodes density is high, the condition is different. When

there is 250 nodes, SDN of SDMAC decreases when R > 200. As the larger the transmission range is,

the more nodes can receive the DRTS and DCTS. However, by ADMAC, a node runs the virtual carrier

sensing and collision avoidance process not only depending on the DNAV but also depending on the angle

between the ongoing transmission direction and the expected transmission direction. Therefore, fewer

nodes have to hold their transmission because of other ongoing transmission than SDMAC.

Figure 11 shows the network throughputs of the four MAC protocols under different nodes density.

Given a total data sending rate by all nodes, the more the senders are, the lower each transmitter’s data

sending load is. As the maximum throughput of every transmitter is limited, when the total data sending

rate is larger than all transmitters’ total maximum throughput, the network throughput does not increase

anymore. In the simulation, we assume that the data sending rate of each node is 100 kbps and the size

of each packet is 1 K bit. Figure 11 shows that the throughput of ADMAC is higher than the other three

MAC protocols. Comparing Figure 11(a) and Figure 11(b), we can know the maximum throughput when

the number of nodes (N) being 150 is larger than that when N = 100. When the number of nodes is

150, the maximum throughput of ADMAC is 114% of the throughput of SDMAC, 123% of DVCS and

130% of DMAC.

Figure 12 shows the network throughputs of ADMAC protocols under different main lobe beamwidths.

As we know, the narrower a directional antenna’s main lobe is, the fewer nodes are blocked by the

transmission and the higher the spatial usage is. In the simulation, R = 100 m and the number of

nodes is 100. Thus, the throughput of ADMAC when beamwidth=20◦ is larger than the throughput

when beamwidth=30◦. The maximum throughput of ADMAC with the beamwidth being 20◦ is about

112.5%of the maximum throughput when the beamwidth is 30◦.

1370 Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6

0 5000 10000 15000 20000 250000

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Total data sending rate (kbps)T

hrou

ghpu

t (kb

ps)

0 5000 10000 15000 20000 250001000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

Total data sending rate (kbps)

Thr

ough

put (

kbps

)

DMACDVCSSDMACADMAC

DMACDVCSSDMACADMAC

×104

Figure 11 The network throughput of four MAC protocols under different nodes densities. (a) N=100, R=100;

(b) N=150, R=100.

0 5000 10000 15000 20000 250001000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

Total data sending rate (kbps)

Thr

ough

put (

kbps

)

Beamwidth=20°Beamwidth=30°

Figure 12 The throughput of ADMAC under different antenna beamwidths.

6 Conclusions

We propose an adaptive directional communication MAC protocol, ADMAC, to make more pairs of nodes

transmit and receive data simultaneously, which increases the network throughput. In order to make a

node send DRTS packet to a receiver correctly, a node needs to learn its neighboring nodes information

and updates it DNAV when it receives packets from others. Before a node starts a transmission, it should

make sure that the outgoing transmission will not interfere with the ongoing transmissions. By changing

the transmission strategy adaptively according to the channel usage, ADMAC protocol can assure nodes

transmitting without interfering with others. A transmitter sends DRTS to a receiver directionally to

decrease the reserved region. On the contrary, the receiver sends an OCTS packet to all directions if the

channel in all directions are idle to avoid the potential transmission interferences later. However, if there

is other ongoing transmission within a receiver’s neighborhood, the receiver runs the collision avoidance

method in Section 4.1. Then it sends a DCTS packet if it does not interfere with the ongoing transmission.

Therefore, more nodes can communicate with each other simultaneously by ADMAC protocol than by

other MAC protocols for directional antennas, such as DMAC, DVCS and SDMAC.

We compare the network throughput of ADMAC with other three MAC protocols for directional

antennas, DMAC, DVCS and SDMAC protocols. The simulation results show that the simultaneous

delivering nodes of ADMAC are more than that of the other three MAC protocols and the maximum

throughput of ADMAC is significantly higher than that of DMAC, DVCS and SDMAC protocols.

Lu X F, et al. Sci China Inf Sci June 2012 Vol. 55 No. 6 1371

Acknowledgements

This work was supported by National Natural Science Foundation of China (Grant Nos. 61100208, 61100205),

Natural Science Foundation of Jiangsu (Grant No. BK2011169). Lio Pietro is supported by the EU FP7 project

RECOGNITION: Relevance and Cognition for Self-Awareness in a Content-Centric Internet.

References

1 Stimson G W. Introduction to Airborne Radar. Mendham: Scitech Pub Inc, 1998

2 Hill J E. Gain of directional antennas. Watkins-Johnson Company Tech-notes, 1976

3 Rappaport T S. Wireless Communications: Principles and Practice. Upper Saddle River: Prentice Hall, 1996

4 Spyropoulos A, Raghavendra C S. Energy efficient communications in ad hoc networks using directional antennas. In:

Proceedings of IEEE INFOCOM, New York, USA, 2002

5 Nasipuri A, Li K, Sappidi U R. Power consumption and throughput in mobile ad hoc networks using directional

antennas. In: Proceedings of IC3N, Miami, USA, 2002

6 Yi S, Pei Y, Kalyanaraman S. On the capacity improvement of ad hoc wireless networks using directional antennas.

In: Proceedings of ACM MobiHoc, Annapolis, USA, 2003

7 Liu B, Liu Z, Towsley D. On the capacity of hybrid wireless networks. In: Proceedings of IEEE INFOCOM, San

Franciso, USA, 2003

8 Yi S, Pei Y, Kalyanaraman S. On the capacity improvement of ad hoc wireless networks using directional antennas.

In: Proceedings of ACM MobiHoc, Annapolis, USA, 2003

9 Yi S, Pei Y, Kalyanaraman S, et al. How is the capacity of ad hoc networks improved with directional antennas?

Wireless Netw, 2007, 13: 635–648

10 Lu X F, Towsley D, Lio P, et al. Minimizing detection probability routing in ad hoc networks using directional antennas.

EURASIP J Wireless Commun Netw, 2009, doi:10.1155/2009/256714

11 Navda V, Subramanian A P, Dhanasekaran K, et al. MobiSteer: using steerable beam directional antenna for vehicular

network access. In: Proceedings of MobiSys’07, New York, USA, 2007

12 IEEE Standards Working Group. Wireless Lan Medium Access Control (MAC) and Physical Layer (phy) Specifications.

1999

13 Shihab E, Cai L, Pan J. A distributed directional-to-directional MAC protocol for asynchronous ad hoc Networks. In:

Proceedings of IEEE GLOBECOM, New Orleans, USA, 2008

14 Bao L, Garcia-Luna-Aceves J J. Receiver-oriented multiple access in ad hoc networks with directional antennas. Wire-

less Netw, 2005, 11: 67–79

15 Janaswamy R. Radiowave propagation and smart antennas for wireless communications. Boston: Kluwer Academic

Publishers, 2001

16 Takai M, Zhou J, Bagrodia R. Performance evaluation of directional adaptive range control in mobile ad hoc networks.

Wireless Netw, 2005, 11: 581–591

17 Li G, Yang L L, Conner W S, et al. Opportunities and challenges for mesh networks using directional antennas. In:

Proceedings of IEEE WiMesh, Santa Clara, USA, 2005

18 Steenstrup M E. Neighbor discovery among mobile nodes equipped with smart antennas. In: Proceedings of Scandi-

navian Workshop on Wireless Adhoc Networks, Stockholm, Sweden, 2003

19 Nasipuri A, Ye S, You J, et al. A MAC protocol for mobile ad hoc networks using directional antennas. In: Proceedings

of IEEE WCNC, Chicago, USA, 2000

20 Ko Y B, Shankarkumar V, Vaidya N H. Medium access control protocols using directional antennas in ad hoc networks.

In: Proceedings of IEEE INFOCOM, Tel-Aviv, Israel, 2000

21 Choudhury R R, Yang X, Ramanathan R, et al. Using directional antennas for medium access control in ad hoc

networks. In: Proceedings of ACM MobiCom, Atlanta, USA, 2002

22 Takai M, Martin J, Ren A, et al. Directional virtual carrier sensing for directional antennas in mobile ad hoc networks.

In: Proceedings of ACM MobiHoc’02, Lausanne, Switzerland, 2002

23 Bao L C, Garcia-Luna-Aceves J J. Transmission scheduling in ad hoc networks with directional antennas. In: Proceed-

ings of ACM MobiCom, Atlanta, USA, 2002

24 Li P, Zhai H Q, Fang Y G. SDMAC: selectively directional MAC protocol for wireless mobile ad hoc networks. Wireless

Netw, 2009, 15: 805–820

25 Ramanathan R, Redi J, Santivanez C, et al. Ad hoc networking with directional antennas: a complete system solution.

IEEE J Selected Areas Commun, 2005, 23:

26 Daeipour E, Bar-Shalom Y, Li X R. Adaptive beam pointing control of phased array radar using an IMM estimator.

In: Proceedings of American Control Conference, Baltimore, USA, 1994