How Conservative IEEE DCF is when using Directional Antenna?

Transcription

1 How Conservative IEEE DCF is when using Directional Antenna? Tamer Nadeem Siemens Corporate Research Princeton, NJ 854 Abstract Directional antennas have been introduced to improve the performance of based wireless networks. Station equipped with directional antennas can beamform data in a specific direction. However, IEEE is developed with omni-directional antennas in mind. Thus, deploying IEEE in a directional antenna environment leads stations to be conservative in blocking their own transmissions in favor of the ongoing transmissions. In this paper we study and analyze the limitation of IEEE networks with directional antennas. We show analytically that a station with directional antenna and using protocol is conservative in terms of assesses channel availability, with as much as 6% of unnecessary blocking assessments. Moreover, by altering the way the accessing its MAC data queue, we show that the unnecessary blocking assessments of a node could be increased up to 9%. Index Terms Wireless Networks, IEEE 82.11, Medium Access Control, Carrier Sense, Directional Antenna, Performance Analysis. I. INTRODUCTION Directional antennas have been introduced to improve the performance of wireless networks [3], [8], [13], [16], [2]. A station equipped with directional antennas can beamform data in a specific direction. The transmitter beamforms the data in the direction of the receiver with diminished interference in the remaining directions. Thus, the network capacity is increased as a consequence of spatial spectrum reuse. The IEEE [1], and carrier sensing protocols in general, was developed with omni-directional antennas in mind. It assumes that all the packets (RTS/CTS/DATA/ACK packets) are transmitted as omni-directional signals that are received by all nearby nodes. Deploying IEEE in a directional antenna environment does not exploit fully the directional antennas characteristics. The main reason is that stations are conservative in blocking their own transmissions in favor of the ongoing transmissions, although their transmissions will not result in interferences with other transmissions. Thus, many modifications (e.g., [3], [4], [6], [17]) were introduced to allow based protocols to exploit the intrinsic features of directional antennas to increase throughput. In this paper, we analyze the performance of IEEE DCF under the use of directional antennas. More specifically, we define two main blocking problems when using the directional antenna. Then we study analytically the limitation Nodes located between R and C Nodes located witihin R Source Destination DIFS Contention Window RTS SIFS CTS SIFS NAV (RTS) NAV (CTS) Data SIFS ACK Fig. 1. IEEE DCF mechanism where EIFS is the Extended Inter Frame-Space period for packets received incorrectly, R is the transmission range, and C is the carrier sense range and conservation of IEEE under these two problems. Later, we verify our analysis through simulation. A. IEEE DCF Mode II. BACKGROUNDS The protocol is developed for wireless networks using omni-directional antennas. The basic MAC method of IEEE 82.11, the DCF, is a Carrier-Sense Multiple Access with Collision Avoidance (CSMA/CA) mechanism. ACK frames are used in the DCF for acknowledging the reception of unicast data frames. The CSMA scheme of the DCF works as follows: Before a station transmits, it must sense the wireless channel to determine if any other stations are transmitting. The channel is assessed as busy if there is detected carrier signal. If the channel is assessed as busy, the station needs to wait until the carrier becomes idle. Then, it waits more for a period known as the Distributed Inter-Frame Space (DIFS). After the DIFS period is passed, the station waits for a random back-off interval and then transmits if the medium is still free. An optional exchange of short Request-To-Send (RTS) and Clear-To-Send (CTS) control frames may precede the actual data frame transmission to reserve the channel. Any station, that overhears either the RTS or the CTS, blocks its own transmissions (if any) to yield to the communication between this sender and its receiver. Each station maintains a timer called the Network Allocation Vector (NAV) which tracks the remaining time of any ongoing data transmission. After a station receives a

2 RTS, CTS, DATA, or ACK frame not destined to itself, it sets its NAV according to the Duration field of the frame. Figure 1 illustrates how stations set their NAVs during a RTS-CTS-DATA-ACK handshake. Checking its NAV before a station attempting to transmit is also known as virtual carrier sensing. If the NAV is not zero, the station needs to block its own transmissions to yield to the ongoing data delivery even if there is no physical signal. In summary, a station blocks its own transmissions if either physical carrier sensing or virtual carrier sensing indicates busy channel. For the rest of the paper, we use carrier sensing to refer to both physical carrier sensing and virtual carrier sensing. B. Directional Antenna Model Directional antennas can transmit data in both omni and directional modes. In omni-directional transmission, the station can transmit with equal power to all directions. In directional transmission, the station directs its energy toward the required direction, called the main lobe, with side and back lobes with lower energy.we adopt the ideally-sectorized pattern [8], [12], that is, there are no side and back lobes. The two-dimensional shape of the adopted directional model is shown in Figure 2. We will discuss the effect of this simplification assumption later. The switched-beam [13] model for directional antenna is used in this paper. In such model, the space of each station is divided into multiple sectors each with predefined fixed beam. Depending on the signal strength and direction, the station chooses one of the possible predetermined beams to transmit or receive data. Figure 3 shows node v with 8 sectors. A station uses both omni and directional modes in receiving ongoing transmission. When the station is idle (not transmitting or receiving), it resides in the omni mode. Once it detects a signal from a certain direction, it starts receiving this signal in omni mode. While it is receiving this signal, it steers the antenna to the direction (or switches to one of the predefined beams) that maximizes the received power. If the node discovers that this transmission is not intended to itself 1, it switches back to omni mode. Otherwise, it switches back to omni mode after it completes the packet reception. Note that the NAV mentioned in the previous subsection is not applicable within such directional transmission of frames. Thus, DNAV, proposed in [3], [17], [18], is used with directional antenna. Unlike NAV, each DNAV is associated with a direction and a width, and multiple DNAVs can be set for a node. A node maintains a unique timer for each DNAV, and updates the direction, width and expiration time of each DNAV every time the physical layer gives newer information about the corresponding ongoing transmission. III. PROBLEM FORMULATION In this section, we define two blocking problems of the IEEE DCF when using directional antennas. 1 Several techniques has been proposed in sensor and ad hoc networks to allow a node to discover whether the ongoing transmission is destined to itself. Among those techniques the use of RTS/CTS, and augmenting PLCP header with additional parameters [1]. Fig. 2. B D E C A Opportunistic MAC example In the original protocol, a station blocks its transmission when it senses a busy carrier. But, if the direction of transmission does not interfere with the ongoing transmissions, this blocking is unnecessary. One way for a station to determine whether its transmission will interfere with these ongoing transmissions is by collecting the locations of transmitters and receivers. Several methods have been proposed in literature allowing a station to acquire its own location, such as using GPS [5], or using RF based localization method [2], [9]. Consider node A that is engaged in a beamforming transmission to the direction of node B as in Figure 2. If Node C wants to beamform data to node D, the running IEEE carrier sense mechanism at node C will block this transmission because of the ongoing transmission between A and B. However, since the ongoing transmission is not destined to C and since the C-D transmission direction would not interfere with A-B transmission as shown in the figure, node C should not block. We refer to this problem as Carrier Sense Blocking (CS Blocking ) problem. The second problem with IEEE DCF happens when the transmission direction of the topmost data item in the node s queue is blocked (due to some ongoing transmission), the station blocks its transmission until that direction becomes free. However, there could exists a subsequent data item in the station s queue in which its transmission s direction does not interfere with any of the ongoing transmissions. Therefore, the station can transmit this data immediately instead of the topmost data. Consider Figure 2 where station A is engaged in a transmission by beamforming data to station B. Assume the top two data items at station C s should be beamformed to stations E and D respectively. Since, C-E transmission is blocked because of the ongoing A-B transmission, IEEE running on station C blocks and wait until C- E direction become free again. However, since the C- D transmission direction would not interfere with A-B transmission, node C should not block, and transmit the seconds data item in its queue to D instead. We refer to this problem as Head-of-Line Blocking (HOL Blocking ) problem. In this paper, we study the CS Blocking and HOL Blocking problems by calculating analytically the potential gain of transmission in the presence of signals in the neighborhood for CS Blocking problem, and presence of blocked direction of the topmost element in the queue for HOL Blocking problem. For both, we verify this potential gain with simulation.

3 IV. RELATED WORKS Previous works address the capacity of wireless networks using directional antennas such as [15], [2]. In [15], authors calculated upper bounds for the capacity gains of using directional antennas. The calculations of the interference based capacity bounds are given for a generic antenna model as well as a real-world antenna model. On the other hands, authors in [2] focused on discovering the lower bounds in capacity improvement that directional antennas can provide. Different RTS/CTS handshake mechanisms with their corresponding analysis for directional antenna are addressed in several works [6], [8], [11], [16] [19] to allow simultaneous transmissions that are disallowed when using only omnidirectional antennas. The effect of using different combination of omni/directional transmissions for one or both of RTS/CTS frames on network performance where studied in term of network throughput [6], [8], [13], [14], [17] [19] or power consumption [11], [16]. All the above mechanisms follow the CSMA mechanism in forcing a node sensing a busy carrier to postpone its transmission although it may not affect the ongoing transmission. DBTMA/DA [7] addressed this problem by avoiding the carrier sense mechanism through splitting a single channel into two sub-channels and used directional busy-tones to accomplish the virtual sensing instead of the physical carrier sense. Using directional transmitting busy tones, it shares the similar feature of the directional RTS frame schemes in that it reserves the network capacity in a finer grain and thus relieves the exposed terminal problem. In the meantime, by using directional receiving busy tones, it realizes a similar functionality of blocking the corresponding antenna element in the direction from which omni-directional CTS frame is received. However, this mechanism require the use of two separate sub-channels which does not follow the standards. Although literature mentioned the inefficiency in IEEE protocol, we are the first, to the best of our knowledge, to analyze the Carrier Sense and Head-of-Line blocking problems in the context of IEEE network using directional antennas. V. ANALYSIS OF BLOCKING PROBABILITIES WITH OPPORTUNISTIC MAC In this section, we (1) study the CS Blocking problem by deriving the probability that a station has an opportunity to transmit directionally despite the presence of transmissions in its vicinity, and (2) study the HOL Blocking problem by deriving the probability that a station has an opportunity to transmit directionally given that the destined sector of the packet at the top of the MAC queue is blocked. Next, we validate the analytical results by simulation and show the potential improvements of opportunistic schemes. In our analysis, we assume two sets of stations: one for transmitters, and another for receivers. Each connection consists of a pair of stations, a transmitter and a receiver. We assume that transmitter, and consequently receivers, are #4 #5 #3 #2 w 2 #6 s 2 v Fig. 3. Blocking analysis where R is the transmission range and C is the carrier sense range uniformly distributed over an area with a density of δ. Each station has a transmission range, R, within which frames sent by the station can be received and decoded, and a carrier sense range C along the directional transmission, which is the range within which transmissions of the station can be detected (channel busy). All stations have the same traffic model, and all data packets are of the same length. Each packet requires transmission time τ, and is randomly destined to a 1-hop neighbor. As mentioned earlier, the space of each transmitter, and consequently receiver, is divided into n sectors (Figure 3 shows a node with 8 sectors). Each transmitter has several ongoing connections in which the corresponding receivers are assumed to be distributed uniformly over m sectors of the n possible sectors, where m n. Each transmitter generates data packets with a rate of 1 T, where T > τ. All transmitters use the same transmission power. r 2 R r 1 #7 w 1 A. Analysis of Carrier Sense Blocking Probability Sometimes, a station unnecessary blocks its transmission, because its carrier sense indicates a busy channel. We say unnecessary because, despite sensing a busy carrier, a station can still transmit without interfering with any of the ongoing transmissions. Consider a scenario (see Figure 3), where station v establishes a connection with station w 2 in sector #4. The standard forces station v to block its transmission once it senses (either physically or virtually) the ongoing transmission between station s 1 and station r 1, or between station s 2 and station r 2. However the directional transmission from station v to station w 2 would not affect any of those ongoing directional transmission, and thus station v should not block its transmission to w 2. In order for station v to avoid interfering with any of the ongoing transmissions of DATA and ACK packets, station v should block its transmission in a specific sector i only if (1) C #1 #8 s 1

4 a sender station s in sector i is transmitting, and station v is in the transmission cone of s (to avoid interfering with ACK reception at s), or 2) a station r in sector i is receiving, and station v is in the reception cone of r (to avoid interfering with DATA reception at r). In case of omni-reception, the condition for v being in the transmission or reception cone is not needed. Assuming that a node v wants to transmit directionally to a sector i with an angular sector η = 2π n, we define the following: P (CS T r ) is the probability that, for all the connections that have their transmitters inside sector i and their corresponding receivers outside sector i, every single transmitter is either not transmitting (1 τ T ), or transmitting and v is outside its transmission cone ( τ 2π η T 2π ). The number of these connections is equal to η 2 C2. The negative term is equal to number of δ η 2 C2 δ n connections that have both their transmitters and receivers in the sector i. P (CS Rcv ) is the probability that, for all the connections that have their transmitters outside sector i and their corresponding receivers inside sector i, every single receiver is either not receiving, or receiving and v is outside its reception cone. The number of these η 2 C2. connections is equal to δ η 2 C2 δ n P (CS T rrcv ) is the probability that, for all the connections that have their transmitters and receivers inside sector i, every single transmitter is either (1) not transmitting (and, thus, the receiver is not receiving), or (2) transmitting and v is outside both the transmission and reception cones of the transmitter and the receiver respectively. The number of these connections is equal to δ n η 2 C2. P (CS Idle ) is the probability that sector i is not blocked and, thus, station v is able to carry out a directional transmission through sector i. P (CS Idle ) equals the multiplication of P (CS T r ), P (CS Rcv ) and P (CS T rrcv ). P (CS Idle ) = P (CS T r ) P (CS Rcv ) P (CS T rrcv ) = [(1 τ T ) + τ 2π η T 2π ](δ δ n ) η 2 C2 [(1 τ T ) + τ 2π η T 2π ](δ δ n ) η 2 C2 2π η 2π η 2π 2π ] δ η n 2 C2 [(1 τ T ) + τ T = [(1 τ 1 T n ) δ(n 1) π n n C2 ] 2 [(1 τ T ) + τ T (n 1 n )2 ] δ π n n C2 (1) In standards, a station v blocks its transmission to sector i if v is in transmission cone of a transmitter in any sector, or v is in the reception cone of a receiver in sector i. Similarly to the above analysis, which has removed due to space limitation, we define P (Std Idle ) to be the probability that sector i is not blocked as follows: P (Std Idle ) = [(1 τ 1 T n ) δ(n 1) n πc 2 ] [(1 τ T ) + τ T (n 1 n )2 ] δ π n C 2 Therefore, P (CS Blocking ), which is the probability that node v blocks unnecessarily, is: (2) P (CS Blocking ) = P (CS Idle ) P (Std Idle ) (3) We verify this analytical model by generating random network topologies and traffic patterns, and studying the blocking probabilities in each case. For constructing each random network, we place the station v at the center of an area of 1m 1m, and distribute transmitter stations uniformly in this area. Each transmitter is paired with a corresponding receiver, which is randomly located within a circular area of radius R that is centered at the transmitter. All packets require transmission time τ, and are generated randomly at a constant rate: one packet every time interval T, where T > τ. The transmission beam width is set at 2π/n where n is number of the neighbor sectors of a node. When station v has a frame to send, it selects randomly a sector to transmit to, and checks whether it can transmit its frame. For the original mechanism, the number of runs in which node v was able to transmit its frame is divided over the total number of transmission attempts to derive P (Std Idle ). Similarly, we can determine P (CS Idle ). We compare P (CS Idle ) and P (Std Idle ), in addition to P (CS Blocking ), to those calculated from the analytical result. Figure 4 plots both the analytical and the simulated curves of P (CS Idle ), P (Std Idle ) and P (CS Blocking ) versus for C = 55m, n = 8, τ/t =.1 and different numbers of stations (thus varying the station density δ). Figure 4 shows that the simulation results closely match the analytical results which validates our analysis. Note that the calculated P (CS Blocking ) is still conservative because we assume, for simplicity, that all stations in the vicinity of v have the freedom of transmission. We do not take into account that some of these stations have to block because of other ongoing transmissions in their vicinities. Accounting for these blocked stations would increase P (CS Blocking ). To analyze how our simplified assumption affects our probability calculations, we relaxed this assumption in the simulation runs as shown in Figure 5. The figure, also, plots P (CS Blocking ) with different packet load values and different sector numbers to show how different parameters affect the P (CS Blocking ). The P (CS Blocking ) plot with conservative assumption, copied from Figure 4, is also included for easy comparison. B. Analysis of Head-of-Line Blocking Probability Following the CS Blocking problem, it happens that the corresponding sector of the packet at top of MAC queue of a station v is blocked, while other sectors corresponding to other packets in the queue are not blocked. To improve network performance, a station should transmit one of the packets corresponding to a non-blocked sector instead of obeying

5 1.8 Analytical P_S Analytical P_O Analytical P_G Simulation P_S Simulation P_O Simulation P_G 1.8 Analytical P_S Analytical P_O Analytical P_G Simulation P_S Simulation P_O Simulation P_G Probability.6.4 Probability Fig. 4. The analytical and simulation values of the probabilities P S=P (Std Idle ), P O=P (CS Idle ), and P G=P (CS Blocking ) Fig. 6. The analytical and simulation values of the probabilities P S=P (Std Idle ), P O=P (HOL Idle ), and P G=P (HOL Blocking ) Probability t =.1, n=8 (from above figure) t =.5, n=8 t =.1, n=8 t =.5, n=8 t =.1, n=4 t =.1, n= Fig. 5. The probability P (CS Blocking ) with different load values t where t = τ/t and different number of sectors n the standard by postponing transmissions until the transmission of packet at the queue s top. Consider the example in Figure 3. Station v has two packets to transmit: the top packet of its queue is to be transmitted to station w 1 in sector #1, and the next is to be transmitted to w 2 in sector #4. Station v is blocked from transmitting to sector #1, because of the transmission from station s 1 to station r 1. In standard 82.11, station v has to postpone all its transmission until sector #1 becomes free. However, since sector #4 is free while sector #1 is blocked, station v should go ahead with its transmission to station w 2 first in order to increase the spatial reuse of the medium. Assuming each transmitter has several connections and the corresponding receivers are distributed uniformly over m sectors of the n possible sectors, we define P (HOL Idle ) to be the probability that at least one sector of the m neighbor sectors is idle (not blocked). Calculating P (HOL Idle ) is tricky, because the blocking probabilities of sectors are not independent, since a transmission may block either one or two sectors. For example, sectors #2 and #6 in Figure 3 are blocked because of the transmission between stations s 2 and r 2. On the other hand, the transmission between stations s 1 and r 1 blocks only sector #1. To simplify the calculation of P (HOL Idle ), we assume the blocking probabilities of sectors are independent. In fact, this simplification will result in calculating the upper bound of P (HOL Idle ) instead. Later, we will show the effect of this simplification and the difference betwen the upper bound and the actual values using simulation. Given this simplification, the upper bound probability of P (HOL Idle ) is calculated as: P (HOL Idle ) = 1 (1 P (CS Idle )) m (4) where m is the number of neighbor sectors and 1 P (CS Idle ) is the probability that a sector is blocked. P (CS Idle ) is calculated in Equation 1. Consequently, the upper bound of P (HOL Blocking ), which is the probability of having at least one of the m neighbor sectors of node v idle, given that the sector corresponding to the topmost packet of the queue is blocked, is: P (HOL Blocking ) = P (HOL Idle ) P (Std Idle ) (5) where P (Std Idle ) is given in Equation 2. Similar to the previous subsection, we verified our analytical results by simulations. In the simulations, whenever v has a frame to send, we select ordered list of m sectors randomly from the n neighbor sectors assuming node v has queue of frames ready to be sent to their destinations in the corresponding m sectors. We study if node v can send any frame from its queue without affecting the ongoing transmissions. The number of runs in which node v is able to transmit is divided over the total number of transmission attempts to derive the probability of P (HOL Idle ). P (Std Idle ), P (HOL Idle ) and P (HOL Blocking ) are compared to the analytical result. Figure 6 plots both the analytical and the simulated values of P (Std Idle ), P (HOL Idle ) and P (HOL Blocking ) for R = 25m, C = 55m, n = 8, m = 4, τ/t =.1, and different numbers of stations (thus varying the station density δ). Although the analytical P (HOL Idle ) and P (HOL Blocking ) are upper bounds, the simulations results closely match those values specially at the peak values, which is our main interest.

6 Probability t =.5, n=8, m=4 t =.1, n=8, m=4 t =.5, n=8, m=4 t =.1, n=4, m=4 t =.1, n=12, m=4 Fig. 7. The probability P (HOL Blocking ) with different load values t where t = τ/t and different number of sectors n while m is fixed to 4 Therefore, the simulation validates our analysis and our simplifications are reasonable. Similar to the analysis of P (CS Blocking ) in the previous section, Figure 7 plots the simulation of P (HOL Blocking ) after relaxing the conservative assumption in calculating P (HOL Blocking ) shown in Figure 6. The figure plots P (HOL Blocking ) with different packet load values and different number of sectors. All previous figures show that the unnecessary blocking probability of a node using standard DCF is large enough in case of using directional antenna (as high as 9% in case of HOL Blocking ). We derived the gain probabilities using the switched-beam antenna model rather than the steerablebeam model for sake of simplicity. The potential gain using switched-beam model is less than that of using steerable-beam model due to the increased flexibility of the latter. Hence, the actual gains are expected to be higher than the one presented here. On the other hand, we assumed in our calculations idealized directional sectors (no side and back lobes) rather than the model adopted in Section II. With idealized sector, interference is ignored and hence the calculated gains are higher than the case with more realistic antenna model. However, since the gain probabilities are significantly high as shown above, this simplification is enough to motivate us to consider modifying the to match and exploit the characteristics of the directional antennas. VI. CONCLUSION AND FUTURE WORKS In this paper, we have defined two major blocking problems of the IEEE DCF when using directional antennas: Carrier Sense Blocking (CS Blocking ), and Headof-Line Blocking (HOL Blocking ) problems. We have shown analytically and by simulation that a station with directional antenna and using protocol is conservative with as much as 6% in case of the CS Blocking problem, and up to 9% for HOL Blocking problem. The study motivates us to design enhancements to the IEEE networks in case of using directional antennas. These enhancements should be able to assist stations to better assess the channel conditions more accurately and allow more concurrent transmissions to take place in presence of detecting busy carrier. REFERENCES [1] IEEE 82.11, Part II: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications. [2] P. Bahl and V. Padmanabhan. RADAR: An In-Building RF-based User Location and Tracking System. In Proceedings of IEEE Infocom 2, Tel-Aviv, Israel, 2. [3] R. Choudhury, X. Yang, R. Ramanathan, and N. Vaidya. Using Directional Antennas for Medium Access Control in Ad Hoc Networks. In Proceedings of the ACM MOBICOM 2, Atlanta, Georgia, USA, September 22. [4] T. ElBatt, T. Anderson, and B. Ryu. Performance Evaluation of Multiple Access Protocols for Ad hoc Networks Using Directional Antennas. In IEEE Wireless Communications and Networking Conference (WCNC), New Orleans, Louisiana, USA, March 23. [5] P. Enge and P. Misra. Special issue on GPS: The Global Positioning System. Proceedings of The IEEE, 87(1), [6] N. Fahmy, T. Todd, and V. Kezys. Ad Hoc Networks with Smart Antennas Using IEEE Based Protocols. In Proceedings of IEEE International Conference on Communications (ICC 22), New York, NY, USA, April 22. [7] Z Huang, C-C. Shen, C. Srisathapornphat, and C. Jaikaeo. A Busy- Tone Based Directional MAC Protocol for Ad Hoc Networks. In IEEE MILCOM, Anaheim, CA, USA, October 22. [8] Y-B. Ko, V. Shankarkumar, and N. Vaidya. Medium Access Control Protocols Using Directional Antennas in Ad Hoc Networks. In Proceedings of IEEE INFOCOM, Tel-Aviv, Israel, March 2. [9] P. Krishnan, A.S. Krishnakumar, W.-H. Ju, C. Mallows, and S. Ganu. A System for LEASE: Location Estimation Assisted by Stationary Emitters for Indoor RF Wireless Networks. In Proceedings of IEEE Infocom 24, Hong Kong, 24. [1] T. Nadeem and L. Ji. Location Aware IEEE for Spatial Reuse Enhancement. IEEE Transactions on Mobile Computing, 6:1171. [11] A. Nasipuri, K. Li, and U. Sappidi. Power Consumption and Throughput in Mobile Ad Hoc Networks using Directional Antennas. In IEEE International Conference on Computer Communication and Networks (ICCCN), Miami, Florida, October 22. [12] A. Nasipuri, S. Ye, J. You, and R. Hiromoto. A MAC Protocol for Mobile Ad Hoc Networks using Directional Antennas. In Proceedings of the IEEE WCNC, Chicago, IL, USA, September 2. [13] R. Ramanathan. On the Performance of Ad Hoc Networks with Beamforming Antennas. In Proceedings of the MobiHOC 1, Long Beach, CA, USA, October 21. [14] M. Sanchez, T. Giles, and J. Zander. Beam Forming Antennas in Multihop Packet Radio. In Proceedings of the Swedish Workshop on Wireless Ad Hoc Networks, March 21. [15] A. Spyropoulo and C. Raghavendra. Capacity Bounds For Ad- Hoc Networks Using Directional Antennas. In Proceedings of IEEE International Conference on Communications (ICC 3), Anchorage, Alaska, USA, May 23. [16] C. Srisathapornphat and C-C. Shen. Energy Consumption Behavior and Performance of Directional Virtual Carrier Sensing Schemes. In Proceedings of the IEEE WCNC, New Orleans, Louisiana, USA, March 23. [17] M. Takai, J. Martin, A. Ren, and R. Bagrodia. Directional Virtual Carrier Sensing for Directional Antennas in Mobile Ad Hoc Networks. In Proceedings of The Third ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc), Lausanne, Switzerland, June 22. [18] M. Takai, J. Martin, A. Ren, and R. Bagrodia. A MAC protocol for full exploitation of Directional Antennas in Ad-hoc Wireless Networks. In Proceedings of The Fourth ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc), Maryland, USA, June 23. [19] K. Xu, M. Gerla, and S. Bae. Effectiveness of RTS/CTS handshake in IEEE based adhoc networks. Ad Hoc Network Journal, Elsevier Science, 1(1), 23. [2] S. Yi, Y. Pei, and S. Kalyanaraman. On the Capacity Improvement of Ad Hoc Wireless Networks Using Directional Antennas. In Proceedings of MobiHhoc 3, Annapolis, Maryland, USA, June 23.