3 Performance Enhancement via Directional Antennas
3.6 Computer Simulations
3.6.4 Effect of Beamforming Null Angle Error
In this section, we discuss the effect of nulling angle error on the throughput performance. The angle error is defined by the difference between the estimated DOA and the real direction of the receiver. We use the topology shown in Figure 3.12.
There are two transmission pairs and the transmission distances are both 100 meters.
Figure 3.16 shows the total throughput of two transmission pairs (A←→B, C←→D) versus the nulling angle error. The transmission maintains a high throughput performance of about 7.6 Mbps within an angle error of 2.2°, because the two transmission pairs do not block each other. However, as the angle error exceeds 2.2°, the throughput decreases drastically but the throughput is still superior to the basic directional antennas mode within the angle error of 3°. Because there exists an acceptable power threshold, CSThreshold (Carrier Sense Threshold), at the physical layer, the throughput looks like a cliff from 2.2° to 3°. When the angle error exceeds 3°, the benefit of the beamforming null is lost, and the 4-null directional antennas mode performs like the basic directional antennas. From this simulation result, it is clear that if we want the M-null directional antennas model to offer better performance, a more accurate neighbor node location identification technique is needed, such as GPS.
3.7 Summary
From the previous chapter, we know that the throughput reduction results from the medium reservation policy in MAC protocol, and the new types of hidden terminal and blocking problems due to improper DMAC protocols. In this chapter, we propose an integrated refinement of physical, MAC and routing protocols with the use of directional antennas. A node equipped with directional antennas has the ability to estimate the DOA of an incoming signal and thus can identify the relative directions of its neighbors. Before transmitting data packets, nodes exchange RTS/CTS packets directionally, and therefore, only those nodes located within the direction of transmission are blocked. Moreover, if a node receives an RTS or a CTS packet from its neighbors, a DNAV is set according to the DOA of that control packet, and the duration as well. With the help of DNAV, a node is blocked only for those directions where RTS/CTS packets come from, and spatial reuse can thus be achieved. Furthermore, to fully address the problems of DMAC protocols, we propose an M-null directional antennas model operating across three layers to form M beamforming nulls to eliminate neighbor interference. The DOA information helps a node to form beamforming null in the M interference directions. Therefore, the M-null directional antennas model can address the problems in existing DMAC protocols and improves the throughput performance significantly.
In Section 3.6, we present some computer simulation results to verify the improvements of the proposed 4-null directional antennas over the omnidirectional and the the basic directional antennas approach. From the simulation results of Figures 3.11 and 3.13, the 4-null directional antennas mode achieves outstanding performance over the basic directional antennas and the IEEE 802.11 omnidirectional antenna. We also discuss the effect of the number of beamforming
nulls; this suggests choosing the most appropriate number of beamforming nulls according to the throughput performance and antenna cost. The result in the simulation of the effect of the nulling angle error shows that the M-null directional antennas mode performs excellent performance within the beamforming null angle error of 3°.
A
B
C
D
E
F G
H J
Figure 3.1: Network capacity is improved via directional antennas. Four sessions is allowed to be held simultaneously without interfering with each other. While in the case of omni-directional communication, most nodes are blocked.
Figure 3.2: Antenna gain pattern of an 8-element circular array.
0º 30º 60º
-30º 100º
-60º 160º
-160º -100º
Figure 3.3 : The antenna gain pattern (dB) of an 9-elements MCMV beamforming can form one main beam towards 0° and eight beamforming nulls in ±30°, ±60°, ±100°, ±160°.
-60º
-160º -100º -30º 0º30º59º 61º100º
Figure 3.4 : The antenna gain pattern (dB) of an 9-element MCMV beamforming can form one main beam towards 0° and eight beamforming nulls in ±30°, −60°, 59˚, 61˚, ±100°, −160°.
Central Element
Passive Parasitical Element
Figure 3.5: A 7-element electronically steerable passive array radiator antenna.
The central element is connected to the main RF radiator. Each passive parasitical element is loaded with a variable reactor.
STEP 1:
A
RTS
B
STEP 2:
A
CTS
B
STEP 3:
A B
DATA
STEP 4:
ACK
B A
Figure 3.6: A modified RTS/CTS exchanging mechanism using directional antennas. The updates of DOA information are achieved by RTS/CTS exchanging. With the latest DOA information, transmission of data packet can be much more reliable.
A
Figure 3.7: Spatial reuse can be achieved by the adoption of DNAV. Two DNAVs, DNAV1 and DNAV3, reserve the wireless medium for nodes B and C. The blank area represents available directions for node A’s transmission.
Neighborhood table
Set transmission direction
& M-null angles (random)
Interference from other
direction?
Memorize it &
replace stale null angle Yes
No
Directional transmission
Figure 3.8: Flowchart of operation of the cross-layer M-null directional antennas system
C
Interference null
Figure 3.9: An example of transmission using the 4-null directional antennas.
Node A equipped with the 4-null directional antennas has packets to transmit to node B. Node A forms a main beam towards B, and four null angles towards nodes C, D, E, and F.
D
Figure 3.10: The 4-node inverted T topology
A B
D Main beam
E F Minor beam
G
0 2 4 6 8 10 12 14 16
Packet Delivery Ratio (%)
Data Generation Rate (Mbps)
Due to hidden terminal
problem
(b)
Figure 3.11: Nulling operation under the 4-Node inverted T topology: (a) Throughput performance versus data generation rate (b) Packet
delivery ratio versus data generation rate
Figure 3.12: The 4-node rhombus topology
0 2 4 6 8 10 12 14 16
Packet Delivery Ratio (%)
Data Generation Rate (Mbps)
(b)
Figure 3.13: Nulling operation under the 4-Node rhombus topology: (a) Throughput performance versus data generation rate (b) Packet delivery ratio versus data generation rate
0 10 20 30 40 50 60 70 80 90 100
Packet Delivery Ratio (%)
(b)
0 10 20 30 40 50 60 70 80 90 100 1.4
1.6 1.8 2 2.2 2.4 2.6 2.8 3
Traffic Load (%)
Average Delay (ns)
4null-directional Basic directional 802.11
Traffic Load (%)
Average Delay (ns)
(c)
Figure 3.14: Nulling operation under the random topology: (a) Throughput performance versus traffic load (b) Packet delivery ratio versus traffic load (c) Average delay of packet transmission versus traffic load
0 10 20 30 40 50 60 70 80 90 100 5
10 15 20 25
Traffic Load (%)
Total Throughput (Mbps)
8null-directional 6null-directional 4null-directional 2null-directional Basic directional
Traffic Load (%)
T o ta l T h ro u g hp ut ( M bps)
Figure 3.15: Effect of the number of beamforming nulls: the throughput performance simulations which discuss on the cases in which nodes are equipped directional antennas with zero, two, four, six, and eight beamforming nulls.
-4 -3 -2 -1 0 1 2 3 4 3.5
4 4.5 5 5.5 6 6.5 7 7.5 8
Angle Error (degree)
Total Throughput (Mbps)
2.2
2.8 2.6
2.5 2.4 2.3
Throughput of basic directional antennas
Due to CSThreshold (carrier sense
threshold)
Total Throughput (Mbps)
Angle Error (degree)
Figure 3.16: Effect of the beamforming null angle error: throughput performance versus angle error
Table 3.1: The routing table of DSDV protocol
Destination Metric Sequence number Next hop Install time
Table 3.2: The update packet of DSDV protocol
Destination Metric Sequence number
Table 3.3: The neighborhood table of modified DSDV protocol
Neighbor Angle Omnidirectional received power P
roInstall time
Table 3.4: Example of routing table update. With the same sequence number, a route with a smaller metric is preferred.
(a) Routing table of node A after receiving the update from node E
Destination Metric Sequence number Next hop Install time A 0 S1012 A T46 D 6 S1000 E T50 C 7 S962 E T50 E 5 S850 E T50
(b) Routing update sent by node B at time T52 Destination Metric Sequence number
B 0 S920 A 1 S1012 C 1 S964 D 2 S1000
(c) Routing table of node A after receiving the update from node B Destination Metric Sequence number Next hop Install time
A 0 S1012 A T46 D 3 S1000 B T52 B 1 S920 B T52 C 2 S964 B T52 E 5 S850 E T50
Chapter 4
DMAC with Cross-Layer Power Control for Wireless Ad Hoc
Networks
Power awareness is an important issue in wireless ad hoc networks. Directional antenna achieves high packet delivery due to high transmission gain, but high transmission gain may be viewed as interference by other transmissions. In Section 2.2.2, we have introduced the blocking problem due to a higher gain. An additional advantage of using directional antennas is the higher gain from the directivity of directional antennas, which can be utilized to reduce the transmission power during the directional transmission. Therefore, we should control power to not only maintain the reliability of data link but also reduce the interference for the other nodes. In this chapter, we will completely illustrate the proposed power control protocol which improves throughput performance, and is demonstrated the improvement of the proposed power control protocol by simulation results.
4.1 Power Scaling in DMAC Transmission
As discussed in Section 2.2, the directional high transmission gain will attack or block the nodes which are out of the omnidirectional transmission range. This case is illustrated by an example in Figure 4.1. Considering a linear topology in Figure 4.1(a), assume that every node is idle in the initial state. In Figure 4.1(b), if node A intends to send a DRTS packet to node B, then node D will be blocked in the direction of node A. In this case, the transmission between node C and D is impossible. The same situation occurred in Figure 4.1(c), node A is blocked by node D, when node D transmits to node C. The Figure 4.1(e) shows that if ad hoc nodes can control transmission power, the two transmissions can survive at the same time.
However, if a node reduces the power of a DATA packet transmission, the reliability of data link will be decreased, which results in low throughput. Therefore, the proposed power control strategy is only to scale the transmission power of control packets DRTS/DCTS/DACK. The DATA packet still transmits with the high directional gain. In the following, we will illustrate the method of the power scaling in the control packets.
Upon node broadcasting periodically, every node can receive the broadcasting signals from neighbor nodes with the omnidirectional received power Pro:
t to ro ro
PG G K
P = d
α (4.1)where Pt is the transmission power of node, Gto is the omnidirectional transmission gain of node’s antenna, Gro is the omnidirectional receiving gain of node’s antenna, d is the distance between transmitter and receiver, K depends on the wavelength, and α is a constant that depends on the propagation conditions.
As the transmitter sends DRTS directionally, the receiver receives DRTS omnidirectionally with the received power Prdo:
t td ro t to ro
rdo
PG G K P G G K
roP = d
α= γ d
α= γ P
(4.2)where Gtd is the directional transmission gain of node’s antenna, and γ is the array gain, which is the ratio of the directional gain Gtd divided the omnidirectional gain Gto.
After discussing the original received power, we want to decide the wanted received power from the scaled transmission power. The simulation engine NCTUns 1.0 defines the carrier sense threshold (CSThreshold). If signal received power is below CSThreshold, the hardware cannot detect this signal. We let Pt'=
β
Pt, whereβ
is the power scaling factor, and is transmission power after scaling. Then, the received power is , and we let = CSThreshold + dB,∆ (dB) is the power increment which is the tolerate range of making sure P that the receiver can detect the signal power after scaling. If we want to send the DRTS packet, then the scaling factor β1 is decided by:
As sending DCTS/DACK, the transmitter and receiver both form directional antennas beamforming, and the directional received power Prd of receiver is:
t td rd t to ro 2
rd
PG G K P G G K
roP = d
α=
γd
γα=
γP
(4.5)Then, the scaling factor
β
2is decided by:2 2 2
Through controlling transmission power by the scaling factors β and 1 β , the 2 receiver node catches the control packets with a small but acceptable power, and the power of the control packets will be not strong interference for the other nodes.
4.2 Operation of the Cross-Layer Power Control Protocol
This section presents the DMAC with the power control protocol which is designed with the cross-layer system architecture in Chapter 3. Figure 4.2 shows the procedure of scaling transmission power on control packets. Firstly, as discussed in Section 3.4, every node broadcasts and receives neighbor node’s broadcast signal, and then records the omnidirectional received power Pro in the neighborhood table.
For an example, assume that transmitter node T has packets to send to receiver node R. According to the neighborhood table, node T gets the received power Pro of node R. Secondly, node T decides the power scaling factor
β
depending on the type of packets. Node T should send DRTS first, and therefore it chooses the power scaling factorAfter node R receiving the DRTS packet, it sends DCTS back to node T with the power scaling factor
=
, so the scaled transmission poweris Pt'=
β
2Pt. Through the power scaling, the small transmission power of these control packets can be accepted by the receiver and does not interfere or block too many of nodes.Since the length of the DATA packet, around 1000 bytes, is much larger than the length of control packets, around 30 bytes, we cannot reduce the transmission power of DATA packets, or the reliability of data link will be decreased seriously, which results in low throughput. In the proposed power control protocol, node T transmits the DATA packet with the full transmission power, so the power scaling factor
β
is 1. Similar to DCTS, node R sends the DACK packet with the power scaling factorβ after receiving the DATA packet successively. 2
4.3 Computer Simulation
The same as Section 3.6, we use the NCTUns 1.0 network simulator [23] to evaluate the performance of the cross-layer power control protocol. In the following simulations, we use the same directional antennas models as indicated in Section 3.6.
We assume that the physical layer at the receiver can accurately estimate the DOA of the received signal and record it in the neighborhood table at routing layer. The packet length is constant and equal to 1017 bytes. We use the two-ray ground propagation model as the path loss model. The channel model uses the Raileigh fading distribution which variance is 10 dB. Each simulation run is conducted for 100 seconds, and each data point is the average of five simulation runs. In the NCTUns simulator, the CSThreshold is 87.57 dB, and we set that the directional gain is 4 and omnidirectional gain is 1, as the same with Section 3.6, so the array gain
γ
is equal to 4. The power increment P∆ is equal to 15 dB. Therefore, the powercontrol factors 1
1 10
10.2574 P
roβ
=
and 21 10
10.25716 P
roβ
=
are obtained by theEquations 4.4 and 4.7. The following sections will represent some simulation results that compare the performance of the 4-null directional antennas using the power control protocol, with the performances of the 4-Null directional antennas not using the power control protocol and the basic directional antennas mode.
4.3.1 Power Control Under 4-Node Topology
The Figure 4.3 shows a 4-node linear topology, which four nodes are arranged to a line. We assume that node A transmits to B and node C transmits to D. As discussed in Section 2.2, the nodes C and D are in the main beam range of node A, and thus they will be interfered or blocked. Although nodes B and C can null the interference through adapting M-null directional antenna model, nodes A and D still attack or block each other, because they are both in the main beam range of each other.
Figure 4.4 shows the simulation results of the two transmissions (A←→B, C←→D) conveying at the same time under linear topology. Figure 4.4(a) shows the throughput performance versus data generation rate. The diamond line, the star line, and the circle line represent the only one transmission (A←→B), two transmissions without the power control scheme, and two transmissions with the power control scheme, respectively. Obviously, two transmissions achieve higher throughput than one transmission, but only a little performance is improved due to the blocking problem. Through the power control on control packets DRTS/DCTS/DACK, we improve 0.9 Mbps throughput, around 22.5%, over the original protocol in the high data generation rate.
Through controlling power on the control packets, the more simultaneous data transmission pairs are allowed, so the network congestion is alleviated. However, the DATA packet is transmitted with the full power, so the DATA packet may be collided, which induces the low packet delivery. Figure 4.4(b) shows the packet delivery ratio as a function of the data generation rate. The power control scheme achieves a lower packet delivery ratio. In spite of the low packet delivery, the cross-layer power control protocol can really improve outstanding throughput performance and save the power consumption.
4.3.2 Power Control Under Random Topology
We now simulate a network consisted of 25 static nodes randomly distributed in a 1000 meters × 1000 meters square area. We have simulated a total of five random scenarios and the results represented Figure 4.5 are the average of their individual results. In addition to simulating the 4-null directional antenna with the power control, we also simulate another power control protocol, the directional medium access protocol with power control (DMAP), which scales power through DRTS/DCTS exchange [12]. DMAP does not consider the routing layer, so it cannot know the received power from neighbor nodes before sending DRTS. This strategy only can scale the power of DCTS through received power of DRTS. Therefore, the DRTS transmitted with the high directional gain will block many nodes, which results in decreasing total network throughput.
Figure 4.5(a) shows the simulation result of the throughput performances as the traffic load increases. As shown in the Figure 4.5(a), the throughput performances of the power control on DRTS, DCTS, and DACK packets, the power control on DCTS and DACK packets, and the 4-null directional antenna without the power control
scheme are approximately the same in the low traffic load, since the nodes are far distant. However, in the high traffic load, the proposed power control protocol enormously outperforms the other protocols. As the power of the DRTS/DCTS/DACK packets is scaled, the throughput has enhancement performance of 7.2 Mbps compared with the 4-null directional antennas not using the power control protocol. The reason is that the network congestion is alleviated. Figure 4.5(b) shows the packet delivery ratio versus traffic load. The proposed power control protocol has the higher packet delivery ratio over the 4-null directional antennas without the power control, since the scaled DCTS and DACK packets have the less probability to interrupt other DATA transmission. However, the proposed power control protocol achieves the lower packet delivery ratio than the power control on DCTS/DACK, since the DRTS with the full directional gain blocks many nodes to avoid collisions. The same performance is represented in Figure 4.5(c), which shows the average delay of the network versus the traffic load. This indicates that the proposed power control protocol excellently enhances the total network throughput over the other protocols, although a little packet delivery ratio is lost.
In addition to considering throughput, we also discuss the issue of the energy efficiency in the following discussion. In order to calculating the power saving, the Equation 4.8 is defined by: packet length, and LA is the DACK packet length. After calculating the average of all simulation results in all the traffic load stages, the proposed power control protocol presents a power saving of 10.71% on average.
4.4 Summary
In this chapter, we proposed the cross layer design of the power control MAC protocol for wireless ad hoc networks. Using the information of routing layer, we can scale the power on the DRTS packet, such that many of the nodes will not be blocked.
As a result, the total throughput performance is enhanced significantly. In addition, our protocol uses the power scaling DCTS and DACK packets to prevent collisions due to a high directional antenna gain. The proposed power control protocol also alleviates the network congestion problem introduced by the RTS/CTS exchanging mechanism. Furthermore, we evaluate the performance of the proposed power control protocol using the NCTUns 1.0 network simulator. The simulation result shows that the proposed power control protocol equipped with the 4-null directional antennas improves the network throughput by 194% over the basic directional antennas and 141% over the 4-null directional antenna without the power control protocol. Finally, in addition to enhancing throughput, the proposed power control protocol on average provides a 10.71% power saving, over the directional antennas protocols. The proposed power control protocol can not only provide an excellent throughput but also reduce the power consumption.
A B C D
Figure 4.1: An example of the benefit of the power control strategy: (a) The initial state of nodes A, B, C, and D is idle. (b) Node A transmits to node B. (c) Node D transmits to node C. (d) Nodes A and D attack and block each other. (e) Through power control, two transmissions can survive at the same time.
Routing
Figure 4.2: Flowchart of operation of the cross-layer power control protocol
A B C D
Figure 4.3: The 4-node linear topology: Node A transmits to node B and node C transmits to node D.