In a wireless ad hoc network, a single transmission channel is shared by all of the stations. The main function of MAC layer is to manage the use of the shared medium. IEEE 802.11 MAC protocol is the most widely used protocol in the current implementation of wireless ad hoc network. This protocol is basically based on the CSMA protocol. In order to solve the hidden node problem, IEEE 802.11 MAC protocol exploits the RTS/CTS exchanging mechanism. The RTS/CTS exchanging mechanism is widely used in wireless ad hoc networks to avoid collisions caused by hidden nodes. Any node that receives an RTS or CTS packet inhibits itself from transmitting. Therefore, data transmission can be completed without the occurrence of collisions.
However, the RTS/CTS exchanging mechanism could lead to blocking problem where a node could be blocked even though there is no nearby node transmitting.
Moreover, the blocking problem may propagate through the network, and the throughput goes down as the load increases. From a network point of view, the blocking problem reveals the tricky point of MAC functionality. On the one hand, MAC protocols must manage the shared medium to ensure successful data transmission, but on the other hand the medium reservation policy induces congestion in the network, like the blocking problem. Directional antennas have been suggested to be used in wireless ad hoc networks to reduce interference outside the intended direction, which increases spatial reuse of the transmission medium.
However, from the previous sections we know that the throughput reduction results from the limitation of medium reservation. Features of directional antenna with
DMAC protocol can alleviate the network congestion.
We have shown that previous research results on DMAC for wireless ad hoc network using directional antenna introduce new types of problems (such as hidden terminal problems, and blocking problem), and the directional antenna models used at network simulation engine (ex. NS2) are too simple to be practical. In the following chapters, we will present our system architecture and computer simulations to show that the proposed scheme overcomes the shortcomings in DMAC and achieves better throughput than IEEE 802.11 and basic directional antenna model with few antenna elements in wireless ad hoc networks.
A B C
Figure 2.1: Hidden node problem: Node C is sending data packet to node B.
Node A is out of radio range of node C, and thus is a hidden node.
Collision occurs at node B when node A initiates transmission.
Figure 2.2: Node B and node C do the RTS/CTS exchanging before data transmission. Node A and node D are blocked by the RTS packet and the CTS packet, respectively.
B C
D
RTS CTS
A
RTS
CTS CTS
RTS
DATA
DATA ACK
ACK
NAV (RTS) NAV (CTS)
SIFS
SIFS SIFS
DIFS
DIFS
DEFER ACCESS BACKOFF
Source
Destination
Others
Figure 2.3: Packet transmission timing based on IEEE 802.11 MAC protocol.
A B
C Blocked
E Blocked
G Blocked
RTS RTS
F
D
Figure 2.4: Blocking problem: Node A and node B are transmitting data packets. Only nodes in the gray area are required to be blocked.
Node E and node G are unnecessarily blocked. This figure also shows how blocking problem propagates.
Omni coverage area
Directional coverage area
Figure 2.5: The sectorized antennas model [4]
Figure 2.6: The switched beam antenna system
A B C D H
E I
F G
Figure 2.7: A topology used to demonstrate the hidden terminal and blocking problems in existing MAC protocols
Chapter 3
Performance Enhancement via Directional antennas
Directional antennas offer tremendous potential for improving the performance of wireless communication systems. Continuing reductions in the cost and size of antenna arrays make it feasible for wireless mobile ad hoc networks. By using directional antennas, radio interference can be reduced effectively, thereby improving the utilization of wireless medium. With directional antennas, simultaneous data transmissions of two or more pairs of nodes located in each other's vicinity may be allowed. However, we have shown that the throughput reduction results from the new types of hidden terminal and blocking problems in existing DMAC protocols.
Therefore the merits of directional antennas may not be fully exploited without the modification. In this chapter, we present an integrated physical, MAC, and routing layers design of new directional antennas model, M-null directional antennas model, to utilize the advantages of directional antennas and address the problems in existing DMAC protocols.
3.1 System Architecture
From the previous chapter, we know that the throughput of a wireless ad hoc network could go to zero as the traffic load increases. The goal is to utilize the features of directional antennas and diminish this situation. Figure 3.1 illustrates a network situation where network capacity is improved by using directional antennas.
Node C is allowed to initiate data transmission to node D while nodes A and B are transmitting. Moreover, although node G is located in the middle of nodes E and F, data transmission between nodes G and J will not interfere with nodes E and F.
However, transmission between nodes G and H is not allowed, because node G is blocked in the direction of data transmission between nodes E and F. In normal situations where nodes are equipped with omni-directional antennas, most of these nodes are blocked. Data transmission between these nodes must take turns by contention. At most two sessions are allowed to be held simultaneously.
In the following discussion, the main idea is to share the knowledge of the wireless medium at the physical layer with higher layers. In standard wireless ad hoc networks, omni-directional antennas are utilized at the physical layer, which provides only information about received power. More useful information, direction of arrival (DOA) can be gained via the use of directional antennas. The DOA information will be shared with higher layers. From the previous chapter, we know that the fundamental resolution to blocking problem is to modify the MAC protocol. The MAC layer protocol should not block all the other neighbors in all directions. Since directional antennas can focus energy in an intended direction, spatial reuse can be achieved. With DOA information available, nodes can be aware of directions of on-going transmissions. As a result, data transmission may be initiated on the premise that no on-going transmissions will be interfered. In order to fully exploit the
advantages of directional antennas, the DOA information can be further utilized at the network layer.
3.2 Physical Layer: Directional antennas
Wireless ad hoc networks are conventionally equipped with omni-directional antennas. However, the technology of directional antennas for wireless mobile communication has received enormous interest in recent years. In this section, we present the implementation of directional antennas at the physical layer.
3.2.1 Features of Directional antennas
Directional antennas have the ability to concentrate the radiation towards the intended direction of transmission or reception. Individual elements on the directional antennas transmit signals omni-directionally, and these signals interfere with each other constructively or destructively. As a result, signal strength increases in one or more directions and eliminates in the others. Consequently, the amount of radiated power to the destined node is reduced, which can largely improve the energy efficiency. However, in the case of omnidirectional antennas, the transmitted power radiates equally well in all directions and only a small percentage of power reaches the destined node. Moreover, because directional antennas have a lower gain outside the intended direction, interference can be minimized. Figure 3.2 shows an example of beam pattern steered by an 8-element circular antenna array [17]. It has a main lobe with about 60° width and gain of 8. The radiated power is concentrated in the range from -30° to 30°, and is suppressed outside the range. In addition to the main
lobe, there are also several side lobes which represent the loss of energy. With more elements on the directional antennas, the increased signal strength in the intended direction can be larger, and the control over beamwidth and direction can be more effectively. The communication area can be extended via the use of directional antennas, and the communication link can benefit more by beamforming at both transmitter and receiver.
3.2.2 Multiply Constrained Minimum Variance (MCMV) Beamforming
Before introducing the proposed M-null directional antennas model, we introduce Multiply Constrained Minimum Variance (MCMV) beamforming [18], which the proposed M-null directional antennas model is based on in this section.
The array data model:
1
( )
D i i
i=
s
θ= ∑ + =
x a n As + n
(D×1) (3.1)where s = [s1, s2, …, sD]T is a transmitted symbol vector, A = [a(
θ
1), a(θ
2),…, a(θ
D)]:D×D, a(
θ
i) is an array steering vector: D×1,θ
i is a DOA of the ith path, and n(k) is a noise vector.The design of an MCMV beamforming involves minimizing the output power subject to the constraints that the desired signal receives a unit gain and the coherent interferers get rejected. Assume that we have information about desired source DOA
θ
1 as well as interference DOAsθ
i, i = 2,…, D. We want to pass desired source distortionlessly while rejecting sources 2 ~ D in a “hard” manner that adds auxiliary constraints to put nulls atθ
i, i = 2,…, D to suppress noise and undetected interferencewith minimum power criterion. Determine the optimum weight vector w by solving the following optimization problem:
2
where Rxx = E{x(k)xH(k)}: D×D is the autocorrelation of x(k).
Solution by Lagrange Multipliers:
1
MCMV beamforming approach can be applied to an array of arbitrary geometry for suppressing coherent and in coherent interference. The optimal weights are generated to form beamfroming nulls in the coherent interferers’ directions of ±30°,
±60°, ±100°, ±160° in Figure 3.3.
If there is an angle error in estimation, the beamforming nulls cannot eliminate coherent interference completely. There are two methods to counteract the angle error.
Firstly, using high-order null constraint is very effective in counteracting the angle error in estimation. Secondly, setting two beamforming null constraints near the direction of main interference can have a wide angle range of beamforming null to tolerate large angle error. For example, as shown in Figure 3.4, we set two beamforming null constraints in 59° and 61°, there is a wider angle range of the beamforming null in 60°.
3.2.3 M-Null Directional antennas Model
Most studies on wireless ad hoc networks with directional antennas have assumed the use of a small, low-cost adaptive antenna which is known as electronically steerable passive array radiator (ESPAR) antenna [19]. As shown in Figure 3.5, the (M+1)-element ESPAR antenna consists of one center element connected to the main radiator and M surrounded passive parasitical elements in a circle. The main radiator exhibits an omni-directional radiation pattern. Each passive parasitical element is loaded with a variable reactor. The antenna pattern is formed according to the bias voltage on the reactors, and thus the reactance values. The ESPAR antenna is capable of forming either omni-pattern or directional pattern. For omni-pattern forming, the bias voltage on each reactor is set equally on condition that the received power is maximized [20]. For directional pattern forming, an optimized set of bias voltage on reactors is obtained such that the received signal power is maximized in the direction of the source.
According to MCMV beamforming in Section 3.2.2, we can use (M+1)-element antennas to build an M-null directional antennas model, which has one main beam,
M-1 side lobes, and M beamforming nulls. For ad hoc networks, a simple circular antenna array is capable of steering a beam through all 360°. M-null directional antennas can concentrate the radiation towards the intended direction of transmission or reception and null interference in specific direction, so it can completely utilize the potential of spatial reuse, and alleviate the hidden terminal problems and blocking problems. In latter simulations, we will show that M-null directional antennas model achieves outstanding performance over the basic directional antennas communications and IEEE 802.11 omnidirectional antennas.
In the following discussion, the antenna system on each node is assumed to be capable of operating in two modes; omni-mode and directional mode. Both the omni and the directional modes can be used to transmit or receive signals. In omni-mode, a node is capable of receiving or transmitting in all directions with a constant gain of Go. A node stays in omni-mode while idle. In directional mode, a node can point its antenna beam towards an intended direction with a gain of Gd which is typically larger than that in omni-mode. Consequently, a node in directional mode has a greater transmission range than in omni-mode. The direction in which the main lobe should be steered for a given transmission is specified to the antenna by the upper layer protocol. When a node is in the omni-directional receiving mode, it is susceptible to interference from all directions. Only when the node has formed a beam to a specific direction, it can avoid the interference from other directions.
3.3 MAC Layer: DMAC
Current MAC protocols, such as IEEE 802.11 standard, do not benefit when using directional antennas, because these protocols have been designed for omnidirectional antennas. To best utilize directional antennas, a suitable MAC protocol must be well designed. The use of RTS/CTS exchanging mechanism is optional in the standard, and is assumed to be used in the following sections. The basic idea of directional MAC (DMAC) protocol is to block only those nodes located within the direction of upcoming data transmission. An intuitive way to achieve this goal is to transmit RTS/CTS packets directionally, which can thus largely reduce the number of blocked nodes. Furthermore, those nodes which received RTS/CTS packets are not blocked in all directions. Through the adoption of directional network allocation vector (DNAV) [6], these nodes can initiate data transmissions in some other directions. In the following, a DMAC protocol based on directional virtual carrier sensing (DVCS) [9] will be introduced.
3.3.1 Neighbor Node Location Identification
In order to steer the antenna beam to an accurate direction in its next hop and send out RTS/CTS packets directionally, the sender needs to know the relative locations of its neighbors. Some related works on DMAC protocol assume that the physical location information may be obtained by using the global positioning system (GPS), ultrasound, or multiple orthogonal channels for control packet transmission. However, these additional resource requirements could make the protocol impractical and unrealistic. For nodes equipped with directional antennas,
neighbor node locations can be obtained by direction of arrival estimation techniques.
Each node caches estimated DOAs from neighbor nodes when it hears any signal, regardless of whether the signal is sent to the node. In a complex environment where lots of scattering, reflection and diffraction could be induced, the result of DOA estimation may not match the physical relative direction which could be obtained by external devices such as GPS. However, the DOA information based on signal strength evaluation may be much more practical than physical location information.
In other words, the DOA is the most effective direction to reach the transmitter with the minimum path loss. The DOA information is updated every time the node receives a new signal from the same neighbor. With DOA information available, RTS/CTS packets can be transmitted in an accurate direction. Therefore, only those nodes located within the direction of upcoming data transmission will be blocked.
The DOA caching mechanism is implemented at routing layer and will be introduced in Section 3.4.2.1.
3.3.2 Modification of RTS/CTS Exchanging Mechanism
This DMAC protocol is an enhancement to the IEEE 802.11 MAC protocol. As mentioned before, two modes of antenna operation are available; omni-directional mode and directional mode. A node listens to the channel omni-directionally when idle. In the DMAC scheme, the RTS, CTS, DATA, and ACK packets are sent directionally. Firstly, as in the IEEE 802.11 protocol, the sender sends out an RTS packet prior to data transmission. The RTS packet is sent directionally to the receiver according to the DOA information. When the receiver receives this RTS packet, it not only updates the DOA information, but also adapts its beam pattern to maximize the received power and locks the pattern for the CTS transmission. Once the sender
receives the CTS packet, it updates the DOA information and adapts its beam pattern for the data transmission as well. This beam pattern adaptation process provides a more reliable data transmission. During the data transmission, beam patterns are locked toward each other for both transmission and reception, and are unlocked after the completion of ACK packet transmission. These locked patterns maximize the signal power at the receiver as long as the channel condition remains the same. Since the period from CTS through ACK transmission is for only a short period of time, the channel response may be assumed to be stable.
Figure 3.6 shows the steps of beam locking and unlocking. Assume that node A has data to be sent to node B. At the first step, node A transmits an RTS packet directionally toward node B according to the last updated DOA from node B.
Although this RTS packet may not be sent in an accurate direction due to node movements, the direction of the upcoming data transmission can be corrected in the following steps. Upon receiving the RTS packet, node B updates its DOA information and locks its beam pattern to this newly derived direction for CTS packet transmission. At the second step, node B sends out a CTS packet toward node A.
Node A updates its DOA information and locks its beam pattern to node B as well. At the third step, node A starts to transmit data packet. These locked beam patterns provide reliable data transmission at both sides. Eventually, data transmission is completed with an ACK packet replied from node B directionally.
3.3.3 Directional Network Allocation Vector (DNAV)
The value of network allocation vector indicates the duration of the ongoing transmission in the vicinity, and the node must reserve the channel for those acting nodes by deferring its own transmission. Directional NAV is an enhancement of NAV,
which reserves the channel only in a certain range of directions. The design of DNAV is to release the medium which is not necessarily reserved, and thus spatial reuse can be achieved. If a node receives an RTS or a CTS packet from its neighbors, a DNAV is set. Each DNAV is tagged with two important values; the direction where the control packet comes from and the duration of the corresponding data transmission.
A node cannot transmit any signals whose direction is in the range of unexpired DNAVs. Another important factor is the width of DNAV, which is based on the beamwidth formed by the directional antennas. The DNAV width of a node must be larger than the beamwidth.
Assume that the width of DNAV is 2w degrees and the beamwidth is 2b degrees.
For a node transmitting safely, it must refer to its DNAV table and the difference between the transmitting direction and all the DNAVs must exceed (w+b) degrees.
In other words, the antenna beam intended to a certain direction must not overlap with any unexpired DNAVs. In the following discussing, we set that the width of DNAV is 45°; the range of DNAV1 is from 315° to 45°; the range of DNAV2 is from 45° to 135°; the range of DNAV3 is from 135° to 225°, and the range of DNAV4 is from 225° to 315°. Figure 3.7 shows a simple example where spatial reuse is achieved.
Node B has data packets to send to node C, and data transmission is initiated after RTS/CTS exchanging. Node A received RTS and CTS packets from nodes B and C, respectively. Two DNAVs with 45° width, DNAV1 and DNAV3, are set upon the reception of these control packets. The numbers in the parentheses represent the relative angles between node A and these two active nodes. If node A has a packet to be sent to node D or node E, it must refer to its DNAV table to check if there is any
Node B has data packets to send to node C, and data transmission is initiated after RTS/CTS exchanging. Node A received RTS and CTS packets from nodes B and C, respectively. Two DNAVs with 45° width, DNAV1 and DNAV3, are set upon the reception of these control packets. The numbers in the parentheses represent the relative angles between node A and these two active nodes. If node A has a packet to be sent to node D or node E, it must refer to its DNAV table to check if there is any