Microsoft Word - Protocol_Speed_Comparison_for_IEEE

Performance Evaluation and Analysis of MANET

Protocols at Varied Speeds

Russell Skaggs-Schellenberg Dr. Nan Wang Daniel Wright

Undergraduate Student: Dept. ECE Professor: Dept. ECE Undergraduate Student: Dept. ECE

California State University, Fresno California State University, Fresno California State University, Fresno

California, USA California, USA California, USA

[email protected] [email protected] [email protected]

Abstract—A Mobile Ad Hoc Network (MANET) is a decentralized

wireless network that does not rely on pre-existing infrastructure.

Instead, it is each node’s responsibility to forward data according

to its specified routing protocol. Although these protocols perform

the same task, their performance in a variety of scenarios differ.

This paper simulates four different routing protocols in NS-3 at a

variety of movement speeds and area sizes, comparing their Packet

Delivery Ratio (PDR) and Average End-To-End Delay (AETED).

The performance results reflect what would be expected if a system

would be implemented in a similar environment. Therefore it is

crucial in choosing a protocol to best suit a system.

Keywords—Reactive, Proactive, Communication Protocol, NS3

Simulation

I. INTRODUCTION

Originally designed for military use, over the past decades

MANET has made its way into the civilian market [1]. MANET

is the mobile implementation of an ad hoc network. Unlike a

centralized network where routing is performed by intermediary

devices, an ad hoc network does not require infrastructure.

Instead, each node is responsible for its peer-to-peer routing to

maintain a self-forming, self-healing network [2]. These nodes

communicate using wireless technology, including Bluetooth,

WiFi, and Zigbee. For nodes not directly connected, routing

protocols are utilized to determine how nodes forward packets

that require multiple hops to reach their destination [3]. A layout

of how a MANET network functions is illustrated in Fig. 1

This paper was inspired by previous work which compared

proactive and reactive protocols in NS-3. [4]. The work focused

on comparing the performance of systems that contained 5, 10,

and 30 nodes, which transmitted 64, 256, and 1024 bit data

packets. The simulation area size and speed remained static.

To accurately simulate a real-world application, the

parameters of the simulation must match the application. It was

predicted that the speed and density of the nodes would best

reflect real-world applications. The question was then asked,

would MANET protocols perform differently when nodes

moved at lower speeds comparatively to higher speeds? Lower

speeds could reflect a scenario where individuals are walking

about, whereas nodes moving at higher speeds could reflect

individuals driving.

Fig. 1. A MANET network.

This paper is broken down into five sections. In the next

section, the three different types of MANET protocols are

discussed. Section III discusses the simulation setup as well as

the performance metrics being evaluated. Section IV breaks up

the gathered results from the simulation between the two metrics

and evaluates the results. The last section summarizes what was

covered, and discusses how this paper will support future work.

II. PROTOCOLS

MANET protocols can be classified into one of three

different categories: proactive, reactive, and hybrid. Proactive

routing protocols maintain a routing table consisting of each

node by periodically propagating update information.

Comparatively, reactive protocols establish a route on-demand

and maintain it until it is no longer used, or no longer accessible.

Lastly, hybrid protocols try to merge the advantages found in

both proactive and reactive protocols [3]. For the focus of this

paper all three will be discussed, though only proactive and

reactive protocols are being evaluated and analyzed. A diagram

of the three categories is shown in Fig.2.

Fig. 2. MANET routing protocols [4].

A. Proactive Protocols

The first proactive protocol is Destination-Sequenced

Distance-Vector Routing (DSDV). This is a table-driven routing

protocol, based off of the Bellman-Ford algorithm. The routing

table consists of each known destination node, as well as which

node the packet should be forwarded to. This assesses how many

hops the packet will need to travel to achieve efficient delivery.

DSDV also utilizes sequence numbers to ensure the fastest route

[5].

Optimized Link State Routing Protocol (OLSR) is the other

proactive protocol being tested. This protocol focuses on

sending out “Hello” messages to its neighbors and topology

control to discover and compute its next hop destination. Once

determined, it uses the shortest hop path to deliver packets [6].

B. Reactive Protocols

The first reactive protocol being tested is Ad Hoc On

Demand Distance Vector Routing (AODV). Like DSDV,

AODV uses sequence numbers to indicate how fresh a route is,

and to avoid loop formation. AODV utilizes three types of

routing messages: Route Request (RREQ), Route Reply

(RREP), and Route Error (RERR). In order to acquire a route to

a destination, AODV floods the network with RREQ. If the

RREQ reaches its destination, then the destination responds by

sending a RREP, otherwise the RREQ would time out and the

last node would respond by sending a RERR [7].

The second protocol is the Dynamic Source Routing (DSR).

DSR is similar to AODV in which it floods route requests to

acquire paths to a destination. However, DSR is different

because it relies on source routing over each node’s routing

table. In source routing, each address between the source and

destination is collected and processed. Once the best route is

processed, the address of each hop is included with the packet

and sent along its designated path [8] [9].

C. Hybrid Protocols

Hybrid protocols are created by merging the advantages of

proactive and reactive protocols. Specific characteristics can be

combined here to tailor a protocol to best suit an application.

Zone Routing Protocol (ZRP) is an example of a hybrid protocol

[10]. In ZRP, zones are formed around each node, which encases

its neighboring nodes. Intra-zone Routing Protocol (IARP) is

used when communicating with neighbors that are within the

zone. At this time a proactive protocol would be used.

Otherwise, when communicating with nodes outside of the

zone, Inter-zone Routing Protocol (IERP) is used which would

consist of a reactive protocol.

D. Modified Protocols

Protocols can also be modified beyond their initial

configuration to improve performance. Mai et al. [11]

implemented a congestion control scheme to AODV to lower

the performance degradation caused by packet congestion. Jhajj

et al. [12] is another example of where AODV was modified. In

this case, the time to live (TTL) factor was used to assist in route

discovery to avoid flooding the network. In both of these cases

AODV was modified to outperform its original protocol.

III. SIMULATION

The simulation software that was used is NS-3, a robust

network simulator. NS-3 was chosen since it supports mobility

and WiFi models as well as MANET routing protocols. It has

excellent documentation and allows for simple network

analysis. NS-3 can also utilize other tools like NetAnim.

NetAnim visualizes the simulation, allowing for the movement

and communication of the nodes to be observed. An example of

a simulation being run through NetAnim is shown in Fig. 3.

Fig. 3. NetAnim visualizing a simulation with a 1000 m

area.

A. Simulation Setup

There are a total of 50 nodes with 10 of them acting as sinks

and 10 sending application data. The data is sent at a constant

rate of 256 bytes per second in the form of four 64 byte UDP

packets. Each simulation runs over a period of 200 seconds and

the application packets start being sent between the 100 and 101

second marks. The nodes use WiFi 802.11b in ad hoc mode and

the transmit power is fixed at 7.5 dBm.

Simulations were run for each protocol (AODV, DSDV,

DSR, OLSR) across several nodes speeds: 0, 5, 10, 15, 20, 25,

and 30. Each set of simulations were run for areas of 500 m

750 m

, and 1000 m

. The complete configuration of the

simulation is tabulated below in Table I.

TABLE

YSTEM

IMULATION

ONFIGURATION

Parameter

Value

Simulator

Network Simulator v3.29

Operating System

Ubuntu 16.04 LTS

Protocols

AODV, DSDV, DSR, OLSR

Simulation Area (m

)

500, 750, 1000

Total amount of nodes 50

Amount of sink nodes 10

Node Speed (m/s)

0, 5, 10, 15, 20, 25, 30

Simulation Time

200 seconds

Channel Type

Wireless Channel

Mac Protocol

802.11b

Data Packet Size

64 bit

Data Packet Type

UDP

Transmit Power

7.5 dBm

B. Metrics

The two metrics that was evaluated during these simulations

were PDR and AETED. Equation (1) was used to determine

PDR for all of the simulations.

(1)

Equation (2) was used to determine AETED for all of the

simulations. The variable n represents the total number of

packets received by the sink nodes. Time Receivedi represents

the simulation time at the moment when a sink node receives a

Packeti and Time Senti represents the time when that packet was

sent.

IV.

SIMULATION RESULTS AND JUSTIFICATION

A. Packet Delivery Ratio Results and Justification

The first set of simulations shown in Fig. 4 compares the

protocols’ PDR across different node speeds in a 500 m

area.

AODV has a higher PDR at higher node speeds, but has a lower

PDR than OLSR and DSR at 5 and 10 m/s. DSDV has the lowest

PDR at each simulated speed.

Fig. 4. PDR vs node speed in a 500 m

area.

In the 750 m

area simulations, shown in Fig. 5, OLSR has the

highest PDR at 0 m/s and AODV being a close second. AODV

has the highest PDR for each other node speed. DSDV has the

lowest PDR for each speed value.

Fig. 5. PDR vs node speed in a 750m2 area.

The last set of simulations have an area of 1000 m

and are

shown in Fig. 6. DSR, OLSR, and AODV have similar PDR at

speed 0 m/s, but AODV tends to have a higher PDR than the

others at higher node speeds.

Fig. 6. PDR vs node speed in a 1000 m

area.

The data collected during the simulations and used for the

PDR line charts is tabulated in Table II.

TABLE

PDR

RESULTS

Area Size

)

Protocols

Speeds (m/s)

0 5 10 15 20 25 30

500

AODV .86 .63 .57 .58 .43 .35 .40

DSDV .79 .39 .27 .24 .27 .12 .14

DSR .99 .70 .48 .47 .41 .33 .30

OLSR .99 .68 .48 .39 .32 .23 .21

750

AODV .97 .50 .51 .41 .38 .35 .27

DSDV .34 .13 .16 .15 .10 .10 .05

DSR .80 .30 .3 .27 .23 .20 .17

OLSR .98 .31 .27 .19 .13 .11 .07

1000

AODV .29 .30 .27 .28 .24 .22 .21

DSDV .10 .05 .03 .07 .12 .08 .03

DSR .30 .14 .11 .22 .20 .22 .20

OLSR .29 .08 .07 .12 .12 .10 .06

B. Average End-to-End Delay Results and Justification

The set of simulations for the 500 m

area in Fig. 7 shows that

DSR has much higher AETED than the other protocols for

speeds greater than 0 m/s with the exception of the 20 m/s

simulation. OLSR has the lowest AETED for speeds below 30

m/s.

Fig. 7. AETED vs node speed in a 500m

area.

Fig. 8. AETED vs node speed in a 750m

area

For the 750 m

area in Fig. 8 and the 1000 m

area in Fig. 9,

DSR has a much higher AETED after 0 m/s and OLSR has the

lowest AETED for each node speed.

Fig 9. PDR vs node speed in a 1000m

area.

The data collected during the simulations and used for the

AETED line charts is tabulated in Table III.

TABLE

III

AETED

RESULTS

CONCLUSION

In this experiment, the PDR and AETED performance metrics

were applied to four different routing protocols. Each protocol

was simulated with nodes traveling at speeds ranging from 0-30

m/s, and area sizes ranging from 500−1000 m

. The data

analyzed will be applied to improve the efficiency of real-world

systems, improvement of protocols, and creation of hybrid

protocols.

CKNOWLEDGMENT

This work was supported by the National Science Foundation

under Grant No. 1816197.

REFERENCES

[1] Bang, Ankur O. and Prabhakar L. Ramteke. “MANET : History ,

Challenges And Applications.”.

[2] AnushkaKhattri, “Introduction of Mobile Ad hoc Network (MANET),”

GeeksforGeeks, 09-Aug-2019. [Online]. Available:

https://www.geeksforgeeks.org/introduction-of-mobile-ad-hoc-

networkmanet/. [Accessed: 17-Nov-2019].

[3] Conti, M. and Giordano, S. (2007). Multihop Ad Hoc Networking: The

Theory. IEEE Communications Magazine, 45(4), pp.78-86.

[4] Y. Bai, Y. Mai and N. Wang, ”Performance comparison and evaluation

of the proactive and reactive routing protocols for MANETs,” 2017

Wireless Telecommunications Symposium (WTS), Chicago, IL, 2017,

pp. 1-5. doi: 10.1109/WTS.2017.7943538

[5] Perkins, Charles and Bhagwat, Pravin. (1999). Highly Dynamic

Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile

Computers. ACM SIGCOMM Computer Communication Review. 24.

10.1145/190314.190336.

[6] En.wikipedia.org. (2019). Optimized Link State Routing Protocol.

[online] Available at:

https://en.wikipedia.org/wiki/Optimized Link State Routing Protocol

[Accessed 17 Nov. 2019].

[7] Jhaveri, Rutvij. (2015). Mobile ad-hoc networking with AODV: A

review. International Journal of Next-Generation Computing. 6. 165191.

[8] Tools.ietf.org. (2019). RFC 4728 - The Dynamic Source Routing

Protocol (DSR) for Mobile Ad Hoc Networks for IPv4. [online]

Available at: https://tools.ietf.org/html/rfc4728 [Accessed 17 Nov.

2019].

[9] En.wikipedia.org. (2019). Dynamic Source Routing. [online] Available

at: https://en.wikipedia.org/wiki/Dynamic Source Routing [Accessed 17

Nov. 2019].

[10] A. Khatkar and Y. Singh, ”Performance Evaluation of Hybrid Routing

Protocols in Mobile Ad Hoc Networks,” 2012 Second International

Conference on Advanced Computing & Communication Technologies,

Rohtak, Haryana, 2012, pp. 542-545. doi: 10.1109/ACCT.2012.86

[11] Y. Mai, F. M. Rodriguez and N. Wang, ”CC-ADOV: An effective

multiple paths congestion control AODV,” 2018 IEEE 8th Annual

Computing and Communication Workshop and Conference (CCWC),

Las Vegas, NV, 2018, pp. 1000-1004. doi:

10.1109/CCWC.2018.8301758

[12] H. Jhajj, R. Datla and N. Wang, ”Design and Implementation of An

Efficient Multipath AODV Routing Algorithm for MANETs,” 2019

IEEE 9th Annual Computing and Communication Workshop and

Conference (CCWC), Las Vegas, NV, USA, 2019, pp. 0527-0531. doi:

10.1109/CCWC.2019.8666607