Scholarly article on topic 'An evaluation of MAC Protocols Running on a MANET Network'

An evaluation of MAC Protocols Running on a MANET Network Academic research paper on "Computer and information sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia Computer Science
OECD Field of science
Keywords
{"Ad hoc networks" / "cognitive radio" / "medium access control" / "multi-channel MAC protocol" / "wireless networks"}

Abstract of research paper on Computer and information sciences, author of scientific article — Luis Fernando Pedraza, Ingrid Patricia Paez, Felipe Forero

Abstract This paper presents a comparative analysis between the mechanisms of media access control IEEE 802.11 and MMAC-CR (Multichannel MAC protocol for Cognitive Radio) in MANETs (Mobile Ad Hoc Networks). The IEEE 802.11 standard allows the use of multiple channels available at the physical layer, but its MAC protocol is designed for a single channel. However, a MAC protocol of a single channel does not work well in a multichannel environment due to the hidden terminal problem. The simulation results show how the MMAC-CR protocol allows a better use of spectral opportunities thereby increasing the throughput of the MANET network.

Academic research paper on topic "An evaluation of MAC Protocols Running on a MANET Network"

Available online at www.sciencedirect.com

SciVerse ScienceDirect

Procedía Computer Science 10 (2012) 86 - 93

The 3rd International Conference on Ambient Systems, Networks and Technologies

An evaluation of MAC protocols running on a MANET

network

Luis Fernando Pedrazaa, Ingrid Patricia Paezb, Felipe Foreroc,a*

aDistrital and Nacional University, Bogota, Colombia bNacional University, Bogota, Colombia cDistrital University, Bogota, Colombia

Abstract

This paper presents a comparative analysis between the mechanisms of media access control IEEE 802.11 and MMAC-CR (Multichannel MAC protocol for Cognitive Radio) in MANETs (Mobile Ad Hoc Networks). The IEEE 802.11 standard allows the use of multiple channels available at the physical layer, but its MAC protocol is designed for a single channel. However, a MAC protocol of a single channel does not work well in a multichannel environment due to the hidden terminal problem. The simulation results show how the MMAC-CR protocol allows a better use of spectral opportunities thereby increasing the throughput of the MANET network.

© 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of [name organizer]

Keywords: Ad hoc networks; cognitive radio; medium access control; multi-channel MAC protocol; wireless networks

1. Introduction

In the last two decades, the experience of users oriented towards collaborative-wise real-time settings has made mobility a dominant aspect of all modern communication technologies. Likewise, current applications need to cope with situations where prompt technological deployment is required within a geographical area with no infrastructure, or situation where it is necessary to set up a communications system over a disaster area due landline collapse. In such circumstances, it is ideal to rely on networks with no infrastructure, such as MANETs, which consist of mobile devices that can connect to each other through shared wireless links and also can establish multi-hop routes whenever the source and destination fall out of their direct transmission range [1]. These networks exhibit some unattractive features, namely a

* Corresponding author. Tel.:+ 57 1 3239300 ext 5814.

E-mail address: lfpedrazam@udistrital.edu.co, lufpedrazama@unal.edu.co.

1877-0509 © 2012 Published by Elsevier Ltd. doi:10.1016/j.procs.2012.06.015

reduced bandwidth that must be shared by all constituent nodes. This problem can be addressed by adopting two different approaches. The first approach, which has been traditionally adopted, corresponds to the IEEE 802.11 standard. The second approach is a novel scheme, namely the MMAC-CR protocol [2], which is employed in Cognitive Radio networks to set up multiple parallel channels, thus ensuring efficient use of spectral opportunities.

The present article discusses a comparative analysis on MANET network behavior when implementing IEEE 802.11 and MMAC-CR as MAC-layer protocols. The fundamentals of these protocols are briefly explained in section II. Section III describes the simulation scenario where the two protocols were evaluated. Results are presented in section IV and are discussed in section V. Conclusions are drawn at the end of this paper in section VI.

2. Preliminaries

This section presents the operation principles of both the IEEE 802.11 and MMAC-CR protocols.

2.1. IEEE 802.11 Distributed Coordination Function

Initially, the purpose of adding multiple channels to the IEEE 802.11 protocol was to improve network performance when having an infrastructure. At present, however, this protocol is also used by ad-hoc networks, using a common channel within the network for communication purposes among all nodes, regardless of whether the nodes are in the transmission range of one of their neighbors or not [3]. For instance, the IEEE 802.11b standard employs DSSS (Direct Sequence Spread Spectrum) together with a physical layer that consists of 14 channels, distant from each other by 5MHz. However, in order to avoid channel overlapping, their frequency distance must be at least 30MHz. Hence, in practice, channels 1, 6 and 11 are used within the 2.4GHz band, as shown in Fig. 1.

DSSS First Set: 3 non-overlapping channels:

Chunni'l 1 Channel 6 Channel II

2.4GHz 2.412GHz 2.437GHz 2.462CHz 2.4S35GIIz

DSSS Second Set: 6 half-overlapping channels

Channel 13 5 7 9!]

i^TYW^ N , I

2.4GHz 2,J12GHi 2.422 2.4J2 2.442 2.452 2,462GHz 2.472GHz 2.4S35GHz

Fig. 1. IEEE 802.11b Channels [4].

IEEE 802.11 operation is based on DCF (Distributed Coordination Function), which is a technique where a node reserves a channel for data transmission by exchanging RTS (Ready to Send) messages and CTS (Clear to Send) messages with the target node. Whenever a node intends to transmit packets to other nodes, first it must send an RTS packet to the destination. The receiving node replies with a CTS packet. Both the RTS packet and the CTS packet include an estimate of the time the channel will be occupied. All the other nodes that hear these packets must postpone their transmissions for a while, as specified in the packets. Hence, every node keeps a variable called NAV (Network Allocation Network), which is a record of the time interval that transmissions should be postponed [5]. This whole process is commonly known as

carrier virtual detection, and allows reserving the area around transmitter and receiver for proper communication, thus avoiding the hidden terminal problem.

Fig. 2 illustrates the operation of IEEE 802.11 DFC. When node B is transmitting a packet to node C, node A hears the RTS packet and sets its NAV up until the end of ACK, likewise, node D hears the CST packet and also sets its NAV to the end of ACK. As soon as transmission is complete, both A and D wait for a time interval called DIFS (DCF Interframe Space) and then contend for the channel. In this example, node B is a hidden terminal with respect to node D. Without channel virtual detection, node D may easily be unaware of the transmission of node B and so begin a packet transmission towards C while B is still transmitting, resulting in packet collision.

Fig. 2. Operation of IEEE 802.11 DCF.

If a node needs to send a packet but notices that the channel is occupied, it chooses a backoff counter of no longer than a time interval called contention window (CW), which corresponds to an independent variable on every node. This variable is reset to the value CWmin when the node is initialized and then again after every successful transmission. After choosing the counter's value, the node waits for the channel to be available and then decrements the count by one after every time slot as long as the channel is not occupied. If the channel is occupied, the node must stop the counter until the channel is available. Because the two nodes can decide on the same backoff counter value, the RTS packet might be dropped due to collision. Since the probability of collision increases with the number of nodes, a transmitter must interpret the absence of a CTS message as a sign of congestion. In this case, the node will double its contention window to reduce the probability of further collisions.

Before transmitting a packet, every node must wait for a short while called inter-frame space, even when the channel is available. There are four different time intervals that activate each packet according to its priority; from shortest to longest these intervals are SIFS (Short Inter-frame Space), PIFS (Point Coordination Function Inter-frame Space), DIFS (Distributed Coordination Function Inter-frame Space) and EIFS (Extended Inter-frame Space). For example, a node waits for a DIFS before transmitting an RTS, but it waits for an SIFS before sending a CTS or an ACK. Thus an ACK may occupy the channel when competing with either an RTS or data packet since the length of an SIFS is shorter than that of a DIFS.

2.2. Cognitive Radio Multi-channel MA C

In this protocol, each Cognitive Radio (CR) keeps two data structures, one representing the Spectral Image of Primary users (SIP) vector and the other representing the Secondary Users Channel Load (SCL) vector. Vector SIP[n] is an estimation of spectrum usage for channel n, and may include the following values [6]:

• SIP[c] = 0 when there is no active Primary User (PU) over channel c.

• SIP[c] = 1 when a Primary User is active over channel c.

• SIP[c] = 2 when there is uncertainty about the presence of PUs.

When a node joins the network, it performs a quick scan over each channel within the ATIM window (Ad Hoc Traffic Indication Message). The result of this channel scan is stored in the SIP vector. After the initial sensing, the values in the vector are updated with the values from the scan. Vector SIP is used to determine whether the network can use a given channel for data transmission. Furthermore, the vector is used to determine whether the node requires scheduling a new scan during data transmission, which becomes necessary when the SIP value of a channel indicates uncertainty.

On the other hand, the SCL vector is used to select the communication channel. This vector comprises the CR expected load per channel. When a node intends to transmit, it chooses the spectral opportunity with the lowest SCL.

Fig. 3 shows the time structure handled by this protocol. Time is divided into fixed-length guide intervals where two phases can be observed, namely the ATIM window and the data window. In the ATIM window, nodes perform a quick scan and also exchange control information. In the data window, data exchange and accurate sensing take place.

J'äst Scan Closed Channel

Control Channel

Fig. 3. Parallel scan and communication over different channels, in this example, nodes 1-5 learn that two Primary nodes (A and B)

are active over channels 1 and 5.

Moreover, in the ATIM window, nodes are aware of the current spectral opportunities throughout the network by listening to the mini-slots of C. The mini-slot protocol is initialized after transmission or reception of the SRP packet (Scan Result Packet) that contains the results of the scan performed over the control channel. This packet is intended to ensure thorough synchronization according to the IEEE 802.11 TFS synchronization function.

If a mini-slot is identified (sensed) as occupied by having a SIP value different from zero, the corresponding channel is excluded from CR communication. Nodes that have packets stored indicate the presence of traffic by sending ATIM frames over the control channel during the ATIM window. In the ATIM frame, a node inserts the preferred channel for transmission, the channel with the lowest SCL and the lowest queue state, for example. Every node that listens to the ATIM frame will update its SCL vector. If the receiving node agrees on the selected channel, it replies with an ATIM-ACK frame. After the ATIM window, the nodes that have exchanged ATIM frames will remain active until the data exchange is complete. The nodes that either did not transmit or did not receive ATIM frames lie idle until the next guide interval.

During the data window, not only data exchange occurs but also the nodes that have a SIP value of uncertainty perform an accurate scan of their corresponding channel. This scan can be run in parallel with communication over another channel. Now the SIP value is updated for this channel. Data exchange follows the normal procedure by performing RTS/CTS exchange according to IEEE 802.11 DCF. An additional feature of this protocol is that the nodes are allowed to stand idle once they complete the exchange within the data window, for example when the transmission queue is empty.

ATIM Window DATA Window

« ti ti

I"' tl ti ti ti ti

¡Skpy M IM HI S DATA

V / ATIM-ACK CTS ACK

3. Simulation Scenario

The impact of Multi-channel MAC on the behavior of a MANET network was evaluated using the simulation tool ns-2.31 [7]. The parameters employed to characterize the network are specified in Table 1.

Table 1. Simulation Parameters

Parameters Value

Area 350X400m

Number of nodes PU 1

Number of nodes CR 6

Mobility Model Random Waypoint

Nodes Speed 5m/s

Pause Time 15s

Network Interface Phy/WirelessPhy

Propagation Model Tworayground

Antenna Type Omnidirectional

Routing Protocol AODV

Transport-layer Protocol TCP

Packet Size 512 bytes

URNG Time 50s

In order to simulate the behaviour of all nodes that constitute the network, the mobility model chosen was RandomWaypoint. In this model, nodes begin at an initial position that is established within the given area and, in the simulation process, they move along a zigzag path at random, as shown in Fig. 4.

Additionally, every node stops moving for a while at any location in order to reduce the effects of sudden changes in direction. The individual movements of the different nodes were generated using another tool, namely BonnMotion v1.5a [8].

(X6.Y6) i / IX2.V 2) (\4,YJ| '' i

\ -- - _ \ i

V", <X3,Y3)

(XI,Yl) <X7,Y7)

Fig. 4. Random Waypoint - Mobility model [9].

To check the multi channel functionality, the CRCN simulator (Cognitive Radio Cognitive Network) proposed in [10] was used. This tool allows using the MMAC-CR protocol, whose operation is divided into two phases. During the first phase, each node sends packets using a preferred reception channel. When there is an available channel, it is selected as the preferred reception channel; otherwise the nodes share the channel with a distant node. During the second phase, the node uses the channel that was selected in phase one to send and receive data. Fig. 5 shows the design structure followed by this protocol.

In terms of routing, MANETs require algorithms that quickly adapt to constant changes in the network topology so as to preserve communication between its constituent nodes. This is the reason why the reactive routing protocol called AODV (Ad Hoc on Demand Distance Vector Routing) was chosen [11].

TCP (Transmission Control Protocol) and FTP (File Transfer Protocol) were the protocols used in the transport and application layers, respectively.

i I.'....,.v:,T

I TCLSaipi ["

ÍIS TCL Nbniiy

Uppef layar

FKuiifig 1-

MAO —1

Natwwk interface

Mobile ncOfl

M Channel U 1 I Chiu-al I I I ChantU? I .. I UvmH n I:

Fig. 5. Design structure ofthe MMAC-CR protocol [10].

4. Results

The first experiment conducted was the MANET implementation using the IEEE 802.11 single-channel protocol. The performance of the network under these conditions is shown in Fig. 6.

TTirMjghpul vs Time

Time (s)

Fig. 6. MANET performance using single-channel IEEE 802.11 protocol.

Then, the behavior of the network using MMAC-CR as the media access control protocol is studied. However, two channels were employed in this experiment. The resulting throughput is shown in Fig. 7.

Figure 7. MANET performance with MMAC-CR protocol and two channels.

Fig. 8 shows throughput values on the network when using three channels with protocols IEEE 802.11 and MMAC-CR, respectively.

Thrwaripytv& Time

Time (s)

Fig. 8. MANET performance when running protocols IEEE 802.11 and MMAC-CR using 3 channels.

Finally, in Fig. 9 is exhibited throughput in the network when four channels are used in MMAC-CR protocol.

Tnioiign^utYS Time

4 -.-i-1-1-1-1---J-1-—I-

I, > II (I tt tj i* J5 « it

Time (3)

Fig. 9. MANET performance using MMAC-CR protocol and fourth channels.

5. Discussion

This section discusses the results of all the experiments conducted. Firstly, when comparing Fig. 6 and 8, it can be observed that the performance of the MANET using IEEE 802.11 together with only one channel is the same as that when using the same protocol (IEEE 802.11) with multiple channels. This result is a natural consequence of the media access control scheme employed by this protocol, which is based on DFC. The design of a MAC protocol that exploits multiple channels when using this protocol is a complicated task, since every IEEE 802.11 device is equipped with a half-duplex transceiver. Although such transceivers are capable of handling channels dynamically, they can only transmit or receive on one channel at a time. Therefore, when a node is listening on a given channel, it is impossible for the same node to listen to the other communications taking place on the other channels, resulting in a multi-channel hidden terminal problem. Hence, the operation of a single-channel MAC protocol, such as IEEE 802.11 DCF, is simply non-optimal in a multi-channel setting, where nodes are able to change channels dynamically [5].

On the other hand, the performance achieved by the MANET when using the MMAC-CR protocol together with two channels is very similar to that of IEEE 802.11 (Fig. 7) since one of these two channels

is being used as Common Control Channel. Fig. 8 shows the benefits of having a third channel; an improvement in throughput values can be observed during the first 35 seconds when using protocol MMAC-CR; this represents a 25% rise compared to IEEE 802.11. This occurs because the third channel compensates for the limited data exchange previously encountered, which resulted from having a control-information dedicated channel. However, this improvement is only relevant when all the nodes are spread around the simulation area and their distance from one another guarantees minimum levels of interference among their transmissions.

On the other hand, after 35s of simulation elapsed time using protocol MMAC-CR, there is a significant reduction in network performance Fig. 8. This type of behavior results from collisions occurring within the network due to node location throughout the simulation area. This shows the impact of choosing an appropriate mobility model on this type of networks.

By including a fourth channel, an average throughput improvement of 50% can be observed when comparing protocol MMAC-CR with IEEE 802.11, as shown in Fig. 9. Only after 44s (simulation time) it is observed that the amount of collisions leads to a considerable throughput decrease when running MMAC-CR.

6. Conclusions

The performance of both the MAC IEEE 802.11 protocol and the MMAC-CR protocol has been studied in the context of MANET networks. In order to do this a script was developed; where the simulation tool NS-2.31, the improvements proposed in [10] and the scenario-generation tool BonnMotion v 1.5a [8] converged.

Simulation results suggest that, as the number of channels increases, the MANET network achieves better performance when running protocol MMAC-CR than when running protocol IEEE 802.11. However, with fewer channels and by applying the present mobility model, protocol IEEE 802.11 exhibits very similar throughput values when compared to those of MMAC-CR.

Basic operation functions of protocol IEEE 802.11 have been fundamental to the development of MMAC-CR protocols, allowing packet transmission coordination, interference reduction, and a CR-wise synchronization of all MAC-associated functions.

Future work should address a comparative study of MANET performance running different multichannel protocols, and applying various mobility models.

References

[1] Berna-Koes M., Nourbakhsh I., and Sycara K., "Communication efficiency in multi-agent systems," in 2004 IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04, vol. 3. IEEE, May 2004, pp. 21292134.

[2] De Domenico A., Calvanese E. S, and Di Benedetto M. G., "A survey on MAC strategies for cognitive radio networks," IEEE Communications Surveys & Tutorials, pp. 1-24, 2010.

[3] Bahl P. and Chandra R., "SSCH: slotted seeded channel hopping for capacity improvement in IEEE 802.11 Ad-Hoc wireless networks," in IEEE 802.11 Ad-Hoc Wireless Networks, in ACM MOBICOM, pp. 216—230, 2004.

[4] "Tech topic 10: License-Exempt wireless applications for public safety," http://transition.fcc.gov/pshs/techtopics/techtopics10.html.

[5] So J. and Vaidya N. H., "Multi-channel MAC for ad hoc networks." ACM Press, 2004, p. 222.

[6] Timmers M., Pollin S., Dejonghe A., Van der Perre L., and Catthoor F., "A distributed multichannel MAC protocol formultihop cognitive radio networks," IEEE Transactions on Vehicular Technology, vol. 59, pp. 446-459, Jan. 2010.

[7] "The network simulator - ns-2," http://isi.edu/nsnam/ns/.

[8] "Informatik 4: BonnMotion," http://net.cs.unibonn. de/wg/cs/applications/bonnmotion/.

[9] Bano N. F., Roppel T., and Gokhale I., "Use ofmobility models for communication in collaborative robotics." IEEE, Mar. 2010, pp. 143-146.

[10] "Cognitive radio cognitive network simulator." http://stuweb.ee.mtu.edu/~ljialian/

[11] Perkins C. E. and Royer E. M., "Ad-hoc on-demand distance vector routing," in Second IEEE Workshop on Mobile Computing Systems and Applications, 1999. Proceedings. WMCSA '99. IEEE, Feb. 1999, pp.90-100.