A MAC Protocol for Directional Hidden Terminal and Minor Lobe ...

0 downloads 0 Views 2MB Size Report
DNAV (Directional NAV), and postpone the commu- nication to the direction of the node which transmitted the RTS or the CTS. At this time, communication to.
A MAC Protocol for Directional Hidden Terminal and Minor Lobe Problems Yuya Takatsuka, Masanori Takata, Masaki Bandai and Takashi Watanabe Graduate School of Informatics Shizuoka University, Japan Email: {takatsuka, takata, bandai, watanabe}@aurum.cs.inf.shizuoka.ac.jp Abstract Recently, several directional MAC (Medium Access Control) protocols have been proposed for wireless ad hoc networks. Directional antennas have significant potentials to improve network performance. However, these directional MAC protocols also have problems which do not exist when using omni-directional antennas. One of the problems is the directional hidden terminal problem. This problem is caused by the difference of the antenna gain between the omni-directional antenna and directional antenna. In addition, a practical antenna has side and back lobes. These minor lobes have non-negligible effects on the interference in network nodes. In this paper, we propose a directional MAC protocol called DMAC-PCDR (Directional MAC with Power Control and Directional Receiving) that mitigates the interference caused by directional hidden terminals and minor lobes. DMAC-PCDR has two features. First, the nodes rotate directionally receiving antenna beams in an idle state. Second, the proposed directional MAC protocol has three access modes and uses these modes depending on the location information. The simulation results show that DMAC-PCDR improves throughput performance.

1. Introduction In recent years, wireless ad hoc networks [1] have attracted a significant amount of attention. Previous studies dealing with wireless ad hoc networks assumed the use of omni-directional antennas. However, IEEE 802.11 [2] is not able to achieve high throughput because a large portion of the network capacity is wasted by extending the wireless media over a large area [3] [4] [5]. To deal with this problem, smart antenna technology may have significant potential [6] [7]. These antennas improve spatial reuse of the wireless channel and extend communication range. Recently, several MAC (Medium Access Control) protocols using smart c 2008 IEEE 978-1-4244-1870-1/08/$25.00 

antennas or directional antennas, typically referred to as directional MAC protocols, have been proposed for wireless ad hoc networks. However, these directional MAC protocols also present new issues which have not been addressed when using omni-directional antennas. Some of the directional MAC protocols use the omni-directional and directional beam for their transmission and reception. Therefore, the difference of the antenna beam gain causes the directional hidden terminal problem. In addition, practical antennas have side and back lobes. These minor lobes have non-negligible effects on the interference among network nodes. In this paper, we explain the directional hidden terminal and minor lobes problems and point out the causes of these problems. We propose a directional MAC protocol called DMAC-PCDR (Directional MAC with Power Control and Directional Receiving) that deals with these problems. The simulation results show that DMAC-PCDR improves throughput performance compared to IEEE 802.11 DCF, DMAC (Directional MAC) and SWAMP-RDR (SWAMP with Rotated Directional Receiving).

2. Related Works IEEE 802.11 DCF (Distributed Coordination Function) [2] is a contention-based MAC protocol of CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) and assumes the use of omnidirectional antennas at the physical layer. The neighboring nodes which receive the RTS (Request To Send) or CTS (Clear To Send) set NAV (Network Allocation Vector) and avoid collision during DATA reception. However, a large portion of the network capacity is wasted using an omni-directional antenna. Therefore, several directional MAC protocols which assume the use of smart antenna have been proposed. Smart antennas improve spatial reuse of the wireless channel and extend communication range.

In DMAC (Directional MAC) [9], a communication pair exchanges RTS/CTS/DATA/ACK, and all frames are transmitted with the directional beam. The receiver receives packets with the directional beam except RTS. The neighboring nodes which receive RTS or CTS set DNAV (Directional NAV), and postpone the communication to the direction of the node which transmitted the RTS or the CTS. At this time, communication to the direction where DNAV is not set is possible. In DMAC, the position information acquisition method required for directional control is not shown. MMAC (Multi-hop RTS MAC) [9] is a protocol which extends DMAC. Multi-hop RTS is used to perform a wider transmission range for communication. SWAMP (Smart Antennas Based Wider-range Access MAC Protocol) [10] based on IEEE 802.11 DCF is composed of two access modes. In the omnidirectional mode, SWAMP exchanges RTS/CTS/SOF (Start Of Frame)/DATA/ACK. SOF is a control frame sent from a sender to indicate the location of the receiver and the start of DATA/ACK exchanges. SOF enables a node 2 hops away from the receiver to determine the location, which is effective in extending the communication range. Each access mode is selected depending on the destination location information. SWAMP-RDR (SWAMP with Rotated Directional Receiving) [8] is a protocol which extends SWAMP for practical smart antennas. Practical smart antennas have side and back lobes. These minor lobes cause new problems which are not caused by ideal antennas. This protocol mitigates the problems by rotating the directional receiving antenna beams and transmitting the NAV request frame. In this protocol, RTR (Ready To Receive) is incorporated into SWAMP. Therefore, SWAMP-RDR has more overhead compared with IEEE 802.11 DCF because of SOF and RTR. On the other hand, DMAC-PCDR does not need these additional control frames. It uses two transmission powers to reduce interference range in contrast to SWAMP-RDR. One of the real smart antennas, ESPAR (Electronically Steerable Passive Array Radiator) antenna [11] which can control the directivity electronically is proposed. The ESPAR antenna is shown in Fig. 1. This antenna is composed of a vertical monopole radiator and six passive radiators. The vertical monopole radiator is used for transmission and receiving. The six passive radiators control the directivity by applying voltage. Some of the antenna beam forms of ESPAR are shown in Fig. 2. The directional beam form has side and back lobes. Our proposed MAC protocol is available for general smart antenna beam forms. In this paper, we use ESPAR antenna beam forms for

12cm

#3 #4

#2

#5

#0

#6

#1

Figure 1. ESPAR antenna. 10

10

5

5

0

0

-5

-5

-10

-10

omni-directional beam form

directional beam form

Figure 2. ESPAR antenna beam forms.

simulations.

3. Directional Hidden Terminal Problem and Side and Back Lobes In this section, we evaluate the basic performance and point out the problems of using practical antenna beam forms.

3.1. Communication Range To explain the directional hidden terminal problem and interference caused by side and back lobes, we make the following assumption. The relationship of the transmitter beam form, receiver beam form and communication range are shown in Table 1. Transmission power control is not used. We assume that the transmitter knows the receiver’s location information.

3.2. Directional Hidden Terminal Problem To show the directional hidden terminal problem, we use a simplified topology as shown in Fig. 3. Ti Table 1. Communication range. Transmitter

Receiver

Range

Omni

Omni

d

Omni

Directional

2d

Directional

Omni

2d

Directional

Directional

4d

A T = T

d

B

d

C

d

D

d

RTS (C→E)

1

2

NAV

DATA (C→E) T = T

d

F

d

3

RTS (A→C)

NAV

Ideal antenna

G S

NAV

CTS (E→C) T = T

E

Directional transmission Directional reception Omnidirectional reception Collision

Figure 3. Directional hidden terminal problem. indicates the time instance (Ti < Ti+1 ). Note that the fan shape in Fig. 3 indicates only the beam form but does not transmission range. For example, at T1 , node C transmits the signal towards node G with directional beam. Since nodes D, E and F wait with omni-directional beam, nodes D and E can receive the signal while node F can not. In this figure, node C is a transmitter and node E is a receiver. At T1 , node C transmits an RTS including its own location information with the directional beam. In an idle state, every node receives packets with the omni-directional beam. Therefore, the communication range is 2d. At T2 , node E transmits a CTS with the directional beam. The neighboring nodes which receive the RTS or the CTS set NAV. Since node A and node B can not receive these packets, they can not set NAV. Node C and node E transmit and receive the CTS, the DATA and the ACK with their antenna main lobe facing each other. At T3 , after node E starts to receive the DATA packet with the directional beam, node A and node B can transmit with the directional beam because they do not set NAV. However, this transmission interferes with the DATA reception of node E. This is an example of the directional hidden terminal problem.

DATA S

DATA S

CTS D

D

D

1. CTS transmission

Practical antenna CTS D

X

2. DATA reception DATA X S

D

X

3. RTS transmission DATA X S

D

X

RTS

S

Collision

RTS X

Figure 4. Interference caused by side and back lobes.

beam because the direction of the packet is not known. Therefore, when node D receives the DATA packet from node S, node X may transmit another packet with the directional beam. In the case of the ideal antenna beam form, this transmission does not interfere with the DATA reception. However, in the case of the practical antenna beam form, this transmission interferes with the DATA reception.

4. Solution of the Directional Hidden Terminal Problem and Influence Caused by Side and Back Lobes In this section, we propose a method to solve the directional hidden terminal problem and interference caused by side and back lobes. The directional hidden terminal problem is caused by the difference of receiving beam gain and transmitting beam gain. One of the solutions to overcome this is to introduce reception with the directional beam. It is assumed that every node can receive a packet with the directional beam.

3.3. Side and Back Lobes Practical antennas have side and back lobes. These minor lobes have non-negligible effects on the interference in network nodes. The interference caused by minor lobes is shown in Fig. 4. The left figure is the case of the ideal antenna beam form, and the right figure is that of the practical antenna beam form. Node S is a transmitter and node D is a receiver. After receiving the RTS, node D transmits a CTS. Then, the neighboring node X can not receive the CTS with an omni-directional beam. In the idle state, every node receives a packet with the omni-directional

4.1. Solution of the Directional Hidden Terminal Problem We explain the effect of the directional hidden terminal problem by introducing reception of the directional beam as shown in Fig. 5. In Fig. 3, the directional hidden terminals are node A and node B. In Fig. 5, every node can receive packets with the directional beam, and node A and node B can receive CTS and set NAV. Therefore, the directional hidden terminal problem is solved.

A T = T

T = T

T = T

d

B

1

2

3

NAV

NAV

d

D

CTS (E→C)

NAV

d

C

NAV

NAV

NAV

NAV

d

E

d

RTS (C→E)

DATA (C→E)

F

G

d

NAV

NAV

NAV

NAV

NAV

NAV

Figure 5. Improvement of the directional hidden terminal problem.

S

S

S

Omni-directional reception Directional reception 1. CTS transmission CTS CTS X D S D X NAV 2. DATA reception DATA D X S D X NAV 3. RTS transmission RTS DATA D X D X S Collision NAV

Figure 6. Improvement of the interference caused by side and back lobes.

4.2. Solution of the Influence Caused by Side and Back Lobes We explain the effect of the interference caused by minor lobes by introducing reception with the directional beam as shown in Fig. 6. The left figure shows reception with the omni-directional beam, and the right figure is the case of reception with the directional beam in an idle state. By introducing reception with the directional beam, the neighboring nodes which can not receive the transmitter’s minor lobe beam with omni-directional beam can receive packets. Therefore, these nodes postpone their transmission and avoid interference with minor lobes.

4.3. Interference range By introducing reception with the directional beam, the interference range is extended. Fig. 5 shows how every node can receive packets with the directional beam, and how RTS is received at node F and node G. These communication protocols reduce spatial reuse of the wireless channel. If node C reduces its transmission

power, there is no interference with node F and node G from the transmission. Thus, node F and node G can communicate with each other while node C is transmitting simultaneously. The method of transmission power control improves spatial reuse of the wireless channel.

5. Proposed MAC Protocol In this section, we propose a directional MAC protocol called DMAC-PCDR (Directional MAC with Power Control and Directional Receiving). DMACPCDR is based on IEEE 802.11 DCF and DMAC. It deals with the influence of the directional hidden terminal problem and side and back lobes by receiving a packet with the directional beam. To receive a packet with the directional beam, each node rotates the directional receiving antenna beams in the idle state. DMAC-PCDR improves spatial reuse of the wireless channel and extends the communication range by transmission power control. The MAC protocol has three access modes, MODE 1, MODE 2 and MODE 3. Each access mode is selected depending on the availability of the receiver location. We assume that every node is equipped with GPS, and the location information required for directional control is obtained in the MAC protocol itself. To obtain the location information, side and back lobes are used subordinately.

5.1. Rotation of the Directionally receiving antenna Beams To receive packets by the directional beam, we propose rotation of the directionally receiving antenna beams. The rotation of the directionally receiving antenna beams solves the directional hidden terminal problem and influence caused by minor lobes. In the idle state, each node rotates the directional receiving antenna beams as shown in Fig. 7. This requires about 200 μs to measure the received signal and to rotate the directional receiving beam 360 degrees if the ESPAR antenna is used [12]. Therefore, in order to enable the receiver to receive the signal, each control packet is transmitted with a preceding tone of about 200 μs. The node which receives the preceding tone stops the rotation and receives packets. Rotation of the directionally receiving antenna beams demonstrates the effects with the CTS that is transmitted by maximum power with a directional beam form which is the same as that of the DATA receiving beam. In Fig. 8, if nodes A, B, X and Y which receive the CTS transmit with the directional beam in the same way as they receive the CTS using same transmission power, then the transmitted packet

1

2

Table 2. Communication range. n

S

3

Transmission Power

4 5

D

A

Figure 7. Rotation of the directionally receiving antenna beams.

B

DNAV DNAV

DNAV

S

D

Receiver

Range

Pt H

Omni

Directional

2d

Pt H

Directional

Directional

4d

Pt L

Directional

Directional

2d

B

d

X

1

DNAV NN

Y

T = T

L

ID

L

C

L(C)

C

L(C)

CTS (P H)

DNAV NN

Figure 8. CTS transmission and set DNAV.

5.2. Transmission Power Control Two transmission power controls are available for DMAC-PCDR. The transmission power levels, transmission beam forms, receiving beam forms, and approximate communication ranges are shown in Table 2. Each node rotates the directional receiving antenna beams in the idle state, so that the receiving beam is the only directional one. The communication range is the maximum range when the transmitter’s and receiver’s antenna main lobes face each other.

5.3. MODE 1 MODE 1 is selected when the transmitter node has no location information of the destination node. The

NN

ID

including

NN

L(C)

Location (C, E)

NN

F

d

d

G

NN ID

L

C

DNAV

C

DNAV NN

L

Receive from back lobe

L(C)

NN

DNAV NN

ID

L

ID

L

ID

L

ID

L

ID

L

ID

L

C

L(C)

C

L(C)

E

L(E)

C

L(C)

C

L(C)

E

L(E)

E

L(E)

FN

FN

interferes with DATA reception. Therefore, the nodes which receive CTS set DNAV. The practical antenna has side and back lobes. If the receiver node receives other packets with these minor lobes, interference occurs. However, use of the pair of rotations of the directionally receiving antenna beams and CTS transmission with a directional beam form which is the same as that of the DATA receiving beam by maximum power solves these minor lobes problems. In this method, for idle node X, if two nodes in its vicinity transmit RTS/CTS control packets simultaneously, then node X may miss one of them. Therefore, node X can not set DNAV to the direction in which node X can not receive the control packet. In this evaluation, we consider these situations.

E

d

DNAV

NN

ID

2

d

including

DNAV

t

DNAV

C Location D (C)

d

RTS (P H) t

T = T

A

Transmitter

FN

ID

L

ID

L

ID

L

E

L(E)

E

L(E)

C

L(C)

DATA (P L) t

T = T

T = T

3

DNAV

4

DNAV

DNAV DNAV

DNAV (P L)

DNAV

DNAV

DNAV

ACK

t

Figure 9. Communication sequence of MODE 1.

sequence of MODE 1 is shown in Fig. 9. The maximum communication range is 2d. In this figure, node C is a transmitter and node E is a receiver. At T1 , node C transmits RTS with the omni-directional beam because node C has no location information of node E. The transmission power is Pt H. The RTS includes location information of node C. At T2 , node E transmits the CTS with the directional beam, and the transmission power is Pt H. The CTS includes location information of node C and node E. The nodes which receive the RTS or the CTS register location information. If the distance of the two nodes is shorter than 2d, it is registered to the NN (Near Node) table, and the other is registered to the FN (Far Node) table. The neighboring nodes which receive the RTS and CTS set DNAV to the direction of node C and node D, respectively. The nodes which receive transmitter’s minor lobes beam can also register location information and set DNAV. After the RTS and CTS handshake, DATA and ACK are transmitted and received with their main lobes face each other with the transmission power Pt L. MODE 1 improves spatial reuse by the directional beam.

A

B

d

C

d

RTS (P L) t

T = T

1

t

T = T

2

NN

ID

L

E

L(E)

inlcuding

d

F

d

including

DNAV

NN

CTS (P H)

D LocationE (C)

d

ID

L(C)

Location (C, E)

C

T = T1

DNAV

L(C)

Receive from back lobe

ID

L

ID

L

ID

L

ID

L

ID

L

ID

L

C

L(C)

C

L(C)

E

L(E)

C

L(C)

C

L(C)

E

L(E)

E

L(E)

NN

NN

NN

DNAV

L(A)

ID L A L(A)

ID L A L(A) FN

CTS (P H)

T = T2

including

ID

L

ID

L

E

L(E)

E

L(E)

C

L(C)

T = T

T = T

3

DNAV

4

DNAV

DNAV DNAV

FN

DNAV (P L)

DNAV

DNAV

DNAV

ACK

ID L E L(E)

ID L A L(A)

ID L A L(A) E L(E)

ID L E L(E)

NN

Figure 10. Communication sequence of MODE 2.

FN

DATA P H

5.4. MODE 2

DNAV

DNAV

DNAV

DNAV

d

G

t

ID L A L(A)

Receive from back lobe DNAV NN

ID L E L(E) FN

ID L A L(A)

)

t

ACK P H (

T = T4

NN

FN

(

T = T3

L(A, E)

Location (A, E) DNAV

t

F

FN

ID L A L(A)

DNAV

ID L E L(E)

d

ID L A L(A)

DNAV

t

E

d

DNAV

NN

NN

L

DATA (P L)

DNAV

NN

D

d

Location (A)

DNAV

t

FN

ID

C

including

FN

ID L E L(E)

NN

FN

FN

d

RTS (P H)

L

DNAV NN

B

d

t

DNAV NN

A

G

NN ID

L

C

d

DNAV

DNAV

DNAV

DNAV

)

Figure 11. Communication sequence of MODE 3. MODE 2 is selected when the transmitter node has location information of the destination node in the NN table. The sequence of MODE 2 is shown in Fig. 10. The maximum communication range is 2d. In MODE 2, all frames are transmitted with the directional beam. The transmission power of RTS, DATA, ACK are Pt L, and that of CTS is Pt H. The location information is distributed in the same way as it is for MODE 1. MODE 2 improves spatial reuse more than MODE 1.

5.5. MODE 3 MODE 3 is selected when the transmitter node has location information of the destination node in the FN table. The sequence of MODE 3 is shown in Fig. 11. The maximum communication range is 4d. In MODE 3, all frames are transmitted with the directional beam, and the transmission power is Pt H. The location information is distributed in the same way as it is for MODE 1 and MODE 2. MODE 3 reduces the number of routing hops by extending communication range.

6. Basic Evaluation In this section, we evaluate the effects on the directional hidden terminal and minor lobes problems of the rotation of the directional receiving antenna beams. In topology 1, we evaluate the effect on the directional

packet A

400m

packet A

B

C 400m (a) Topology 1 packet D

400m

B E

C

D

200m 200m

400m 400m (b) Topology 2 Figure 12. Topology.

hidden terminal problem. In topology 2, we evaluate the effect on the minor lobes problem. The relationship of the transmitter beam form, receiver beam form and communication range are shown in Table 1. In this section, we assume the communication range d to be 250m. We evaluate two protocols, the original DMAC and the Rotation DMAC which used rotation of the directional receiving antenna beams.

500 450 s)p b 400 K (t 350 up hg 300 uo 250 rh T 200 et ag 150 reg g 100 A 50 0

1200 Original DMAC Rotation DMAC

0

10

20 30 40 50 60 Arrival Rate per Node λ (packets/sec)

s)p 1000 b K( tu 800 ph gu roh 600 T et ag 400 reg g 200 A 70

0

Original DMAC Rotation DMAC

0

10

20 30 40 50 60 Arrival Rate per Node λ (packets/sec)

70

Figure 13. Basic evaluation (Straight topology).

Figure 14. Basic evaluation (Cross topology).

6.1. Topology 1 (Effect on the directional hidden terminal problem)

B at minor lobes. The simulation result in topology 2 is shown in Fig. 14. In the original DMAC, the reception of node C is interfered with by the directional hidden terminals. And node B is interfered with by node D and node E by minor lobes. On the other hand, in Rotation DMAC, node A can receive CTS from node C. And node D and node E can receive CTS from node B from the minor lobes. Then, node A, node C and node E postpone their transmission and avoid interference. Therefore, the minor lobes problem is also solved by rotating the directional receiving antenna beams.

In topology 1, we evaluate the effect on the directional hidden terminal problem of the rotation of the directional receiving antenna beams. The four nodes are arranged in a straight line as shown in Fig. 12 (a). Packets are delivered from node A to node D on three hops. In this topology, when the node B and node C pair is communicating or when the node C and node D pair is communicating, node A or node B is the directional hidden terminal and interferes with the reception of node C and node D, respectively. The simulation result in topology 1 is shown in Fig. 13. In the original DMAC, the reception of nodes C and D has interference from the directional hidden terminals, node A and node B, respectively. On the other hand, in Rotation DMAC, node A and node B can receive CTS from node C and node D, respectively. Then, node A and node B postpone their transmission and avoid interference. Therefore, the directional hidden terminal problem is solved by rotating the directional receiving antenna beams.

6.2. Topology 2 (Effect on the minor lobes problem) In topology 2, we evaluate the effect on the minor lobes problem. The five nodes are arranged at the crossing point shown in Fig. 12 (b). Packets are delivered from node A to node C on two hops and from node D to node E on one hop. In topology 2, when the node B and node C pair is communicating, node A is the directional hidden terminal, and it interferes with the reception of node C. And the transmission from node D and node E interfere with the reception of node

7. Advanced Evaluation In this section, we evaluate the performance of DMAC-PCDR. The evaluated protocols are IEEE 802.11 DCF, the original DMAC, SWAMP-RDR and DMAC-PCDR.

7.1. Simulation Parameters We make the following assumptions. A hundred nodes are arranged at random in a square area with dimensions of 1500 m. The nodes are fixed. Packets arrive at every node according to the Poisson distribution with a mean value of λ (packets/sec). The destination node for each packet is chosen at random within two hop communication neighbors. The packet size is 1460 bytes and the data rate is 2 Mbps. For the directional antenna beam forms, we use ESPAR antenna beam forms. In IEEE 802.11 DCF, when the transmitter transmits with the omni-directional beam and the receiver receives with the omni-directional beam, the communication range is 250m. In DMAC, when the transmitter transmits with the directional beam and the receiver

9 s)p 8 b7 M (t up 6 hg 5 uo rh 4 T tea 3 ge rg 2 g A1 0

14

s)p b M (t up hg uo rh T tea ge rg g A

12

10

IEEE 802.11 DCF DMAC SWAMP-RDR DMAC-PCDR

8

6

IEEE 802.11 DCF DMAC SWAMP-RDR DMAC-PCDR

4

2

0

0

5

10 15 20 25 Arrival Rate per Node λ (packets/s)

30

0

1000

2000

3000

4000

5000

Data Size (bytes)

Figure 15. Aggregate Throughput versus the packet arrival rate.

Figure 16. Aggregate Throughput versus the data size.

receives the omni-directional beam, the communication range is 250m. In SWAMP-RDR and DMACPCDR, when the transmitter transmits with the omnidirectional beam and the receiver receives with the directional beam, the communication range is 250m.

The aggregate throughput versus the data size is shown in Fig. 16. Poisson distribution λ is fixed to 10 (packets/sec). DMAC-PCDR has higher throughput as the data size is set to be large. DMAC can be easily interfered with by the neighbor node transmission during DATA reception. The communication period is longer due to the large data size. Consequently, the long period is wasted which is caused by collision which decreases throughput of the DMAC. On the other hand, DMAC-PCDR reduces influence with collisions by rotation with the directional receive antenna beams and by CTS which is transmitted by maximum power with a directional beam form. Therefore, the proposed protocol can better transmit a large size packet without collisions compared with DMAC. SWAMP-RDR also reduces influence with collisions. However, SWAMP-RDR has two additional control frames. These are overheads. Therefore, DMAC-PCDR has higher throughput than SWAMP-RDR.

7.2. Throughput Evaluation The aggregate throughput versus the packet arrival rate is shown in Fig. 15. Compared with IEEE 802.11 DCF and DMAC, DMAC-PCDR improves throughput by about 3 and 1.5 Mbps, respectively. In DMAC, the transmitter transmits with the directional beam and the receiver receives with the directional beam expecting RTS reception. Therefore, DMAC improves spatial reuse of the wireless channel and throughput more than IEEE 802.11 DCF does. However, DMAC is interfered with by the directional hidden terminal problem and side and back lobes. On the other hand, DMAC-PCDR reduces interference by rotating the directional receiving antenna beams. DMAC-PCDR reduces the number of routing hops by extending communication range and the interference area by transmission power control. Therefore, DMAC-PCDR improves throughput more than DMAC does. SWAMP-RDR also reduces interference and routing hops. However, SWAMP-RDR has two additional control frames. Therefore, DMACPCDR improves throughput more than SWAMP-RDR does. When the receiver is near, SWAMP-RDR uses omni-directional RTS, omni-directional SOF and directional RTR. On the other hand, DMAC-PCDR uses directional RTS and does not use SOF and RTR. Therefore, when the density of the node is high, it is considered that DMAC-PCDR is more advantageous than SWAMP-RDR.

7.3. One Hop Delay The one hop delay versus the packet arrival rate is shown in Fig. 17. In this evaluation, delay means the average time of one hop communication. In DMAC, the delay is decreased compared with IEEE 802.11 DCF. This occurs because DMAC allows the node to transmit packets toward the direction where DNAV is not set. Therefore, in DMAC, many communication pairs communicate simultaneously. When the packet arrival rate is low, the delay of DMAC-PCDR is decreased compared with DMAC. In DMAC-PCDR, a preceding tone of about 200 μs is needed for the RTS and CTS transmissions, and the number of nodes which set DNAV is more than DMAC because every node receives packets with the directional beam in the

0.12

has three access modes and uses these modes depending on the location information. To obtain the location information, side and back lobes are used subordinately. The simulated results show that the throughput of DMAC-PCDR is improved compared with that of IEEE 802.11, DMAC and SWAMP-RDR. Our proposed MAC protocol is available for general smart antenna beam forms as well as ESPAR antenna beam forms.

0.1 ) s (

0.08

y a l e D

0.06

p o H 1

0.04

IEEE 802.11 DCF DMAC SWAMP-RDR DMAC-PCDR

0.02 0

0

5

10 15 20 25 Arrival Rate per Node λ (packets/s)

30

Figure 17. One Hop Delay.

idle state. However, the probability of setting DNAV is low, and the transmitter can transmit the packet right away without interference when the packet arrival rate is low. On the other hand, in DMAC, a preceding tone is not needed, but the probability of interference is higher than it is for DMAC-PCDR. Therefore, the delay of DMAC-PCDR is decreased compared with DMAC when the packet arrival rate is low. When the packet arrival rate is high, the delay of DMACPCDR is increased compared with DMAC. In DMACPCDR, the probability of setting DNAV is higher when the packet arrival rate is high. Many transmitters postpone their transmission more than DMAC does for avoiding interference which causes high delay. Therefore, the delay of DMAC-PCDR is increased compared with DMAC when the packet arrival rate is high. DMAC-PCDR reduces the number of hops by extending communication range more than DMAC does. As a result, DMAC-PCDR improves throughput more than DMAC though the delay is large. DMACPCDR improves the delay more than SWAMP-RDR. This occurs because DMAC-PCDR reduces control frames more than SWAMP-RDR does.

8. Conclusion In this paper, at first, we have explained the directional hidden terminal problem. This problem is caused by the difference of receiving beam gain and transmitting beam gain. We have briefly explained the solution methods. After analyzing the problems, we have proposed the directional MAC protocol called DMAC-PCDR. DMAC-PCDR deals with the directional hidden terminal problem and side and back lobes problems by rotating the directional receiving beams and transmission power control. DMAC-PCDR

References [1] R. Jurdak, C. V. Lopes and P. Baldi, “A Survey, Classification and Comparative Analysis of Medium Access Control Protocols for Ad Hoc Networks,” IEEE Communications Surveys and Tutorials, vol.6, no.1, First Quarter 2004. [2] ANSI/IEEE Std 802.11, “Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications,” 1999. [3] S. Xu and T. Saadawi, “Does the IEEE 802.11 MAC Protocol Work Well in Multihop Wireless AdHoc Networks?,” IEEE Commun. Mag., vol.39, no.6, pp.130137, June 2001. [4] J. Li, C. Blake, D. S. J. D. Couto, H. I. Lee, R. Morris, “Capacity of Ad Hoc Wireless Networks,” Proc. ACM/IEEE Mobile Computing and Networks (MobiCom), 2001. [5] S. Wu, Y. Tseng, and J. Sheu, “Intelligent Medium Access for Mobile Ad Hoc Networks with Busy Tones and Power Control,” IEEE Journal on Selected Areas in Communications, vol.18, no.9, Sep.2000. [6] P. H. Lehne and M. Pettersen, “An Overview of Smart Antenna Technology for Mobile Communications Systems,” IEEE Communications Surveys and Tutorials, vol.2, no.4, Fourth Quarter 1999. [7] J. H. Winters, “Smart Antennas for Wireless Systems,” IEEE Personal Communications, Feb. 1998. [8] Y. Takatsuka, K. Nagashima, M. Takata, M. Bandai, T. Watanabe, “A Directional MAC Protocol for Practical Smart Antennas,” Proc. IEEE Global Communications Conference (GLOBECOM), 2006. [9] R. R. Choudhury, S. Yang, R. Ramanathan and N. H. Vaidya, “Using Directional Antennas for Medium Access Control in Ad Hoc Networks,” Proc. ACM Mobile Computing and Networking (MobiCom), pp.5970, Sep. 2002. [10] M. Takata, K. Nagashima, T. Watanabe, “A Dual Access Mode MAC Protocol for Ad Hoc Networks Using Smart Antennas,” IEEE International Conference on Communications (ICC), pp.4182-4186, June 2004.

[11] S. Chandran (Ed), “Adaptive Antenna Arrays Trends and Applications,” Springer, pp.184-204, 2004. [12] T. Ueda, S. Tanaka, D. Saha, S. Roy, and S. Bandyopadhyay, “Location-Aware Power-Efficient Directional MAC Protocol in Ad Hoc Networks Using Directional Antenna,” IEICE Trans. Commun., vol.E88-B, No.3, pp.1169-1181, March 2005.