The Impact of Transmission Opportunity (TXOP) on the

Performance of Priority based Contention based Scheduling

Strategies in Multi-hop Mesh Networks

Sajid M. Sheikh, Riaan Wolhuter and Herman A. Engelbrecht

Department of Electrical and Electronic Engineering, University of Stellenbosch,

Private Bag X1, Matieland, 7602, South Africa

Keywords: EDCA, Fairness, MAC, Wireless Mesh Networks, Scheduling, IEEE802.11e, Priority Scheduling.

Abstract: Wireless Mesh Networks (WMNs) face multiple problems. An increase in the number of hops for packets to

reach the destination results in an increase in contention for the medium which also results in an increase in

the collision rates. The enhanced distributed channel access (EDCA) mechanism was developed to provide

differentiated services to data with different priority levels in the IEEE 802.11e standard. The EDCA is a

distributed, contention-based channel access mechanism of the hybrid coordination function (HCF) which

results in an unfairness problem where higher priority data can starve lower priority data. We adopt the

EDCA architecture for heterogeneous data in telemetry and IoT applications to address these problems of

EDCA in multi-hop mesh networks. An adaptive weighted round robin (AWRR) scheduling strategy has

been proposed and tested on multi-hop networks in our previous work. With the AWRR strategy, although

packet loss is reduced, the end-to-end delay increases with high and medium priority data compared to

EDCA in WMNs. In this paper we investigate the effect of the Transmission Opportunity (TXOP) bursting

on the global performance in a WMN through setting up simulations in OMNeT++ using the INETMANET

framework. Simulation results have shown that using TXOP–bursting in the priority based scheduling

which follows the concept of schedule before backup helps reduce packet loss as well as reduce the end-to-

end delay. TXOP further optimizes the performance of AWRR.

1 INTRODUCTION

Wireless Mesh Networks (WMNs) have been

viewed as a promising technology for low cost

deployments for telemetry networks in rural areas as

well as for extending network coverage compared to

other solutions such as fiber optic, cellular networks,

Wi-Max or VSATs (iDirect 2009; Hammond and

Paul, 2006). WMNs have also attracted deployment

in many applications due to their self healing

properties where data can use an alternate route to

send data to the destination in the event when a link

becoming faulty (Akyildiz et al., 2005). Despite

these advantages, WMNs experience some

performance limitations. As stated in (Akyildiz et

al., 2005; Sheikh et al., 2015 and Pathak and Dutta,

2011) some of the main challenges are (1) the drop

in throughput with an increase in the number of hops

for data to reach the destination, (2) an increase in

the contention for the medium which results in an

increase in the collision rates and (3) a starvation

problem (fairness) affecting lower priority data with

the use of the enhanced distributed channel access

(EDCA) scheduling mechanism. Data that traverses

multiple hops to reach the destination usually

experience more contention to access the medium

compared to nodes closer to the destination (Denko

and Obaidat, 2009). The unfairness problem takes

place as the nodes closer to the receiver are given a

higher chance to transmit their data than those

progressively further way (Denko and Obaidat,

2009). EDCA has an internal node contention

mechanism such that when two data packets try to

transmit data on the medium at the same time, the

higher priority data is given access to the medium

and the lower priority data behaves as if a collision

occurred on the medium and exponentially increases

it’s contention window size resulting in starvation

for lower priority data (Chen, 2011 and Telenor et

al., 2005).

In many implementations and research, WMNs

usually use the EDCA scheduling strategy which

enhances the popular carrier sense multiple access

Sheikh, S., Wolhuter, R. and Engelbrecht, H.

The Impact of Transmission Opportunity (TXOP) on the Performance of Priority based Contention based Scheduling Strategies in Multi-hop Mesh Networks.

DOI: 10.5220/0005949201130120

In Proceedings of the 13th International Joint Conference on e-Business and Telecommunications (ICETE 2016) - Volume 6: WINSYS, pages 113-120

ISBN: 978-989-758-196-0

113

with collision avoidance (CSMA/CA) for data of

different priority levels. The original CSMA/CA was

designed for wireless local area networks (WLANs)

based on signal-hop networks (Denko and Obaidat,

2009). To address the fairness and contention

increase problems in WMNs, a novel distributed

contention based mechanism called adaptive

weighted round robin (AWRR) in (Sheikh et al.,

2015) was proposed. Weighted round robin (WRR)

scheduling had been applied before for WiMax

scheduling as in (Guesmi and Maaloul, 2013), for

single hop WLANs IEEE802.11 based networks in

(Kuppa and Prakash, 2004; Farn and Chang, 2005

and Lee et al., 2005, but not for multi-hop WMNs.

With this proposed scheme, packet loss is reduced as

well as improvement in fairness to address starvation

as the internal collisions mechanism is removed.

With this strategy different nodes might be

transmitting data from a different queue as there are

guaranteed slots for different queues which results in

an increase in end-to-end delay for high and medium

priority data and a reduction for low priority data.

EDCA contains a contention free period known

as Transmission Opportunity (TXOP). When one of

the priority classes gain access to the channel to send

data, multiple frames can be transmitted in the

duration of the TXOP without the need to sense the

channel again and perform the back-off period (Inan

et al., 2007; Suzuki et al., 2006 and Min et al., 2008.

This condition is only valid as long as the duration

does not exceed the TXOP limit set. Each packet

transmission in the TXOP duration is separated by

Short Inter-frame Spacing (SIFS). If the TXOP limit

is set to 0, only one frame can be sent when the

priority class gains access to the channel (Inan et al.,

2007). Transmitting of multiple frames during the

TXOP is also referred to bursting as many packets

are transmitted continuously (Suzuki et al., 2006).

In the study in (Hu et al., 2012), it is shown that

TXOP with Quality of Service (QoS) differentiation

helps improve the system performance. The AWRR

strategy has been tested without focus on the

integration of TXOP. In this paper, we carry out a

comparative analysis of EDCA and AWRR with and

without the implementation of TXOP bursting in

multi-hop mesh networks. In (Reddy et al., 2007),

the proposed strategy focuses on dynamically

changing the TXOP limit values. In (Reddy et al.,

2007), an Adaptive-TXOP (A-TXOP) is proposed

where the TXOP interval is dynamically adjusted

based on the packets in the queue. The TXOP in our

understanding also contributes to a form of

unfairness as it allows multiple consecutive packets

to be transmitted of the same priority class.

Therefore, if high priority data is starving lower

priority data; it is expected than TXOP will result in

further unfairness.

EDCA was mainly designed for multimedia

applications such as voice and video which can

tolerate small amounts of packet loss but require less

end-to-end delay (Gao et al., 2005). There are many

non-delay sensitive applications that require a high

degree of reliability (less packet loss) QoS over

delay. This is to say that they can tolerate slightly

more delay provided it is within the tolerable ranges.

Examples of these applications are smart rural

applications such as smart grid, smart buildings,

smart farming and smart health (Sheikh et al., 2015).

These applications carry heterogeneous type of data

having different priority levels running on the same

communication network. In (Sheikh et al., 2015), the

requirements of these applications have been

classified into three categories, namely high,

medium and low priority. For EDCA to be used in

these applications to carry data of different priority

levels, it will have to be able to provide a high

degree of reliability as well as provide end-to-end

delay within tolerable ranges.

The novel contribution of this paper is that we

investigate the impact of TXOP (also know as

bursting) on the performance of EDCA and AWRR.

With EDCA packets from the different queues

concurrently try to access the medium by performing

their back-off countdown in parallel. With AWRR,

the packets only perform back-off after they have

been selected for transmissions. The rest of this

paper is organised as follows. Section 2 presents an

overview of EDCA which is a contention based

strategy in the IEEE802.11e standard. In section 3,

we present an overview of the AWRR scheduling

strategy. In section 4, an overview of the simulation

setup and performance metrics used for the analysis

are presented. In Section 5, the results and presented

and in section 6, the paper is concluded.

2 PRIORITY SCHEDULING IN

THE IEEE802.11E STANDARD

In this section we present a brief overview of the

EDCA scheduling strategy which is used as the

baseline strategy in this paper. EDCA is an

enhancement of CSMA/CA which is widely used in

many WMNs implementations.

In the IEEE802.11 standard, the Medium Access

Control (MAC) layer has two access mechanisms,

namely the contention based method called

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

114

distributed coordination function (DCF) and the

contention–free method called point coordination

function (PCF) (Maamar et al., 2011). The PCF is

the infrastructure based technique while DCF is the

distributed technique where the devices content for

the medium (Kaveh Pahlavan, 2002). The PCF is

used less in WMNs implementations due to the

difficulty of achieving time synchronisation globally

within a network.

With DCF, data of different priority is treated

equally and in a first in first out (FIFO) transmission

queue scheduling strategy. To provide differentiated

services the IEEE802.11e standard was proposed.

The IEEE 802.11e standard is based on both the

centrally-controlled and contention based medium

access mechanism (Kaveh Pahlavan, 2002). The

hybrid coordination function (HCF) is used in IEEE

802.11e which combines the aspects of DCF and

PCF with enhanced QoS mechanisms to provide

service differentiation providing both distributed and

centrally controlled channel access mechanisms.

EDCA is the distributed, contention-based channel

access mechanism of HCF (Poonguzhali, 2014).

Table 1: EDCA parameters.

AC Traffic Type

AIFS

No.

min

max

TXOP limit

802.11a PHY

TXOP limit

802.11b PHY

AC[3] Background 7 31 1023 0 0

AC[2] Best Effort 3 31 1023 0 0

AC[1] Video 2 15 31 3.008ms 6.016ms

AC[0] Voice 2 7 15 1.504ms 3.264ms



Figure 1: Reference EDCA implementation model for

IEEE802.11e (Sheikh et al., 2015).

Figure 2: TXOP Limit.

EDCA consists of more than one queue for data of

different priority levels known as access categories

(ACs). Each one of these ACs has specific

parameters associated with it as shown in table 1.

These parameters are designed such that high

priority data have smaller values than lower priority,

giving the higher priority data a higher chance to

access the channel (Pan et al., 2009).

Data is mapped at the MAC layer into the

corresponding AC. EDCA introduces a new

interframe spacing called Arbitration IFS (AIFS).

AIFS is the minimum time period for which the

medium must be sensed idle before an Enhanced

Distributed Channel Access Function (EDCAF) may

start transmission or back-off. The period is

depended on the AIFSN, CWmin and CWmax

values as shown in table 1. The higher priority ACs

have smaller CWmin and CWmax values compared

to lower priority ACs (Poonguzhali, 2014). For each

of the ACs, the corresponding AIFSN, CW values

and TXOP limit values are also shown in table 1.

Figure 1 shows the implementation scheduling

structure of EDCA. If any queue has data, data is

scheduled after sensing the medium to be idle for the

AIFS period and CW backoff period depending on

the priority class.

TXOP is a time interval during which a station

can send multiple frames one after the other

separated by a SIFS period as shown in figure 2. In

the EDCA standard, the TXOP limit is set to 3.264

ms for voice data and 6.016 ms for video data if the

IEEE 802.11b standard is used and to 1.504ms for

voice data and 3.008ms for video data if the IEEE

802.11a standard is used. For data, the TXOP–

bursting is set to 0 (Suzuki et al., 2006). These

TXOP limit values have been setup to suit voice and

video QoS required and packet sizes.

3 ADAPTIVE WEIGHTED

ROUND ROBIN (AWRR)

SCHEDULING STRATEGY

In this section we briefly explain how the AWRR

scheduling strategy works. To this AWRR strategy

we integrate a TXOP mechanism and test this

strategy with different TXOP limit values for data.

With AWRR, information from the header on the

type of application the packet is coming from is used

to classify and place the frames in the different

priority queues at the MAC layer. Weights are

assigned to the different priority queues. In our case

we have assigned 50% for high priority data, 30%

The Impact of Transmission Opportunity (TXOP) on the Performance of Priority based Contention based Scheduling Strategies in

Multi-hop Mesh Networks

115

for medium priority data and 20% for low priority

data. Based on these weights, we assign 10 slots to

these queues. The numbers of slots assigned to the

different queues can be changed. They are

application dependent and are

dependent on how much

transmission probability chance you want to assign to the

different queues (Sheikh et al., 2015). Table 2 presents the

slots assigned to the different queues depending on which

queues have

data. The weights only apply if all the

queues have data. Figure 3 shows the complete

overview of the AWRR strategy. The frame only

gets transmitted after performing the AIFS and back-

off according to the priority data (Sheikh et al.,

2015).

With the AWRR scheduling strategy, only one

frame gets scheduled at a time as compared to

EDCA where the frames from the different queues

contend for the medium. After the scheduling

process, the AIFS period and back off are carried out

before transmission on the medium takes place.

There is no internal contention mechanism in

AWRR. To perform the investigations in the paper,

the TXOP limit mechanism is added after the back-

off period.

4 SIMULATION SETUP

Many telemetry and Internet of Things (IoT)

applications such as smart grid, home-automation,

health-care monitoring are characterized as

consisting of heterogeneous data in the network.

These heterogeneous data have different priority

levels depending on the applications. To investigate

the effect of TXOP on EDCA and the AWRR

scheduling strategy for data, simulations were setup

in OMNeT++ using the INETMANET framework.

Table 3 presents the simulation parameters. The

standard IEEE802.11e model with AC[0] for high

priority data, AC[1] for medium priority data and

AC[2] for low priority data was used. The traffic

types are heterogeneous with different priority

levels. The two ray ground propagation model was

used to represent the physical environment as the

main focus of testing these strategies are for rural

smart applications. Usually the numbers of obstacles

or buildings are less in most rural areas in Africa.

The two ray model was used as in rural areas,

predominantly these two rays exist, i.e. direct rays

and the reflected rays due to less development in

rural areas. Shadowing models are more suitable for

more developed areas with more obstacles. User

Data Protocol (UDP) packets at the transport layer

having sizes of 512 bytes were used as UDP

Table 2: Queue slots assigned (Sheikh et al., 2015).

Does queue have data? Adaptive Queue Weights Assigned

High

Medium Low High Medium Low

No Yes 0 0 1

Yes No 0 1 0

Yes

No No 1 0 0

Yes Yes 0 3 2

Yes

No Yes 5 0 2

Yes

Yes No 5 3 0

Yes

Yes Yes 5 3 2



Figure 3: AWRR scheduling strategy (Sheikh et al., 2015).

applications such as Trivial File Transfer Protocol

(TFTP) and Domain Name Systems (DNS) use a

default packet size of 512 bytes. UDP was used as it

does not establish connections between the source

and destinations (connectionless) and also there is no

retransmission of lost packets [40]. The use of UDP

packets helps to determine the unreliability of the

network at the lower layers through packet loss

measures. On the other hand, Transmission Control

Protocol (TCP) is connection oriented and also

feedback is received for delivered packets

(Xylomenos and Polyzos, 1999).

A 5x5 square grid topology was used for the

investigation with measurements being done at the

source and sink nodes being the furthest apart in the

network as shown in figure 4. Source 1 and Sink 1

are other nodes also in communication to have a

scenario with data links also in communication. The

transmission range of each node is set such that each

node can only communicate with its adjacent nodes.

Square grid topologies present higher contention

levels with a high number of neighbouring nodes.

This helps to access the performance in extreme

cases.

Five test cases with different TXOP limit values

were setup with EDCA and AWRR as shown in

table 4. For each of these test cases, the performance

was tested over different data transmission rates with

constant bit rate (CBR) data as shown in table 5.

Each test with each test case and each data

transmission date rate was repeated 10 times with

different seed numbers generated by the random

number generator utility in OMNeT++. Each seed

number was 10000 apart. The errors bars show the

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

116

95% confidence level.

The performance metrics used in this paper are:

1. Number of Collisions: There are two types of

collisions that can take place with the EDCA.

These are internal and external collisions.

Internal collisions take place in the node if

EDCA is used. In AWRR no internal collisions

take place. External collisions take place on the

channel when packets collide physically. The

total number of collisions is calculated as:

TotalnumberofCollisionspersecond



Externalcollisions  Internalcollisions

simulationtimein seconds

(1)

2. End-to-end Delay: This is the average time delay

by a packet to arrive at the destination from the

source.

3. Percentage Packet Loss (%): This is the measure

of the percentage of packets lost from the source

to the destination. This value was measured at

the destination as (Periyasamy and Karthikeyan,

2014):

Packe

Loss %





#o

Packetstransmitted  # o

packetsrecieved



∗ 100

#o

Packetstransmitted

(2)

4. Jain Fairness Index (JFI): It measures how fairly

or unfairly the resources are shared among the

nodes. Equation 3 presents the JFI value where x

is the normalized throughput of station i, and n is

the number of flows in the WMN. A JFI of 1

indicates absolute fairness and a JFI of 0 absolute

unfair resource distribution (Deng and Han,

2009). In our case n=3 as we investigate the

fairness for 3 data classes namely for high,

medium and low priority classes.









,



,



,….,







∑















∑













(3)

0 









,



,



,….,





1

5 RESULTS

Figure 5 presents the number of collisions that took

place in the network with EDCA and AWRR with

the different TXOP test cases. In the TXOP tests, the

value for the low priority data TXOP is kept smaller

than the TXOP limit values for high and medium

priority data as normally the higher priority data

need to be transmitted with higher importance. For

AWRR in TXOP test cases 1 to 5, the number of

collisions per millisecond starts at 4.12 per ms with

case 1 and reduces to 3.87 per ms in case 4. With

EDCA, the numbers of collisions stay approximately

the same, despite using TXOP, due to the internal

Table 3: Simulation Parameters.

Network Setup

Topology type 5 by 5 Grid Topology

Terrain Area 2.2km x 2.2km = 4.84km

IEEE Standard IEEE 802.11g

Propagation Model Two Ray Ground Model

Routing Protocol OLSR

Data rate 54Mbits/s

Transport Protocol UDP Packets

Packet Size 512bytes

Table 4: TXOP test cases.

Case 1 Case 2 Case 3 Case 4 Case 5

High Priority Data

No TXOP 1ms 2ms 3ms 4ms

Medium Priority Data

No TXOP 0.5ms 1.5ms 2.5ms 3.5ms

Low Priority Data

No TXOP 0 1ms 1ms 1ms

Table 5: Data transmission test cases.

High Priority

Data

(Packets/sec)

Medium

Priority Data

(Packets/sec)

Low Priority

Data

(Packets /sec)

Data Case 1

50 50 50

Data Case 2

50 50 100

Data Case 3

50 100 50

Data Case 4

50 100 100

Data Case 5

100 50 50

Data Case 6

100 50 100

Data Case 7

100 100 50

Data Case 8

100 100 100

Figure 4: Test Topology.

collision mechanism being present which starves

lower priority data. The higher priority data have a

smaller collision window range and hence the

chances of collisions are high. With AWRR, the

reduction in collision with higher TXOP values is

observed since when a higher priority data gains

access to the medium, more higher priority data can

be transmitted without having to contend for the

medium as they have a higher collision possibility.

AWRR also allows more packets from other classes

to be transmitted on the medium compared to

EDCA.

For brevity, the consolidated average packet loss

over all the data transmission data rates with the

different TXOP test cases are presented in figures 6

to 8. From figure 6, we can observe that for high

priority data using EDCA, there is a packet loss of

50.8% in case 1 (No TXOP) and this reduces to

The Impact of Transmission Opportunity (TXOP) on the Performance of Priority based Contention based Scheduling Strategies in

Multi-hop Mesh Networks

117

50.4% until TXOP case 3. For high priority data

using AWRR, there is a packet loss of 44.8% in case

1 (No TXOP) and this reduces to 37.6% until TXOP

case 3. Therefore, a further packet loss reduction of

7% with AWRR for high priority data is observed.

From figure 7, we can observe that for medium

priority data using EDCA, there is a packet loss of

45.5% in case 1 and this reduces to 44.5% until

TXOP case 3. For medium priority data using

AWRR, there is a packet loss of 42.5% in case 1 and

this reduces to 36.6% until TXOP case 3. Therefore,

a further packet loss reduction of 1% for EDCA and

5.9% with AWRR for medium priority data is

observed. From figure 8, we can observe that for low

priority data using EDCA, there is a packet loss of

40.5% in case 1 and this reduces to 36.9% until

TXOP case 3. For low priority data using AWRR,

there is a packet loss of 39.1% in case 1 (No TXOP)

and this reduces to 33% until TXOP case 3.

Therefore, a further packet loss reduction of 1% for

EDCA and 6.1% with AWRR for low priority data

is observed. It is observed that increasing TXOP

beyond case 3 does not affect packet loss any further

for either EDCA or AWRR.

It is observed that TXOP does not significantly

affect packet loss in EDCA, but does significantly

reduce the packet loss in AWRR. The TXOP limit

allows multiple frames to be transmitted in this

TXOP duration without the need to contend for the

medium for each frame in the queue. The TXOP

only comes into play when more than one frame is

present in the queue. During the TXOP the medium

is sensed as being busy by the other nodes

contending for the medium and therefore, this results

in fewer collisions for the extra frames transmitted

during this period. The packet loss is reduced until a

point and further increasing the TXOP has no effect

on the performance. This is as a result that there are

no packets queued up further for the longer TXOP to

come into play.

With AWRR, the back-off countdown is started

only after scheduling a packet, while with EDCA,

the packets in the different queues in a node perform

the count down simultaneously. After a transmission

takes place, the queues that were already counting

down for back-off start from where they left off,

while with AWRR a new back-off is started. This

Figure 5: Number of Collisions.

Figure 6: Average packet loss for hig

priority data in all TXOP test cases.

Figure 7: Average packet loss for mediu

priority data in all TXOP test cases.

Figure 8: Average packet loss for lo

priority data in all TXOP test cases.

Figure 9: Average end-to-end delay for

high priority data in all TXOP test

cases.

Figure 10: Average end-to-end delay for

medium priority data in all TXOP test

cases.

Figure 11: Average end-to-end delay

for low priority data in all TXOP test

cases.

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

118

increases the chances of collision on the medium

with EDCA.

The consolidated average end-to-end over all the

data transmission test cases for the different TXOP

test cases are shown in figures 9 to 11. From figure 9

we observe an end-to-end delay of 5.3ms for high

priority data and 56ms for AWRR in case 1. The

end-to-end delay drops to 5ms with EDCA until case

3 and drops to 34ms for AWRR. A reduction in the

end-to-end delay of 23ms is observed with AWRR.

From figure 10 for case 1, we observe an end-to-end

delay of 8.2ms with EDCA and 34ms with AWRR.

The end-to-end delay drops to 8ms with EDCA until

case 3 and drops to 25ms for AWRR. A reduction in

the end-to-end delay of 9ms is observed with

AWRR. From figure 11 for case 1 it is observed that

EDCA has an end-to-end of 69ms and 33ms for

AWRR. The end-to-end delay drops to 49ms with

EDCA until case 3 and drops to 26ms for AWRR. A

reduction in the end-to-end delay of 7ms is observed

with AWRR and 20ms with EDCA. Higher end-to-

end delay reductions are observed with AWRR as

AWRR does not starve lower priority data and gives

a higher chance for data from other classes to be

transmitted as mentioned earlier.

The Jain’s fairness index values are presented in

figure 12. No significant change in fairness is

observed. Since the internal collision mechanism is

absent in AWRR, AWRR has a higher fairness than

EDCA. Application of TXOP with AWRR is

expected to improve fairness at higher loads as

lower priority data are given a fair opportunity and

are not being starved.

It must be noted that these strategies have a retry

limit of 7. The packets that collide increase the end-

to-end delay. These results are for heavy load

scenarios. With low loads, performance is better.

Figure 12: Jain’s Fairness Index in all TXOP test cases.

6 CONCLUSIONS

The problem with contention based scheduling

strategies is that they require monitoring of the

channel before data can be transmitted. The

advantage of TXOP is that multiple packets from the

same queue can be transmitted without the need of

continuously performing the contention period. With

AWRR using TXOP, a reduction in collisions has

been shown as the channel is sensed as being busy

by the other nodes during the TXOP period and the

other packets within the TXOP period of the same

data class can successful transmit. Retransmission of

collided packets waste channel bandwidth and

reduce the overall performance of the network.

Bandwidth is a critical factor in rural telemetry

networks.

In this study, we have observed that with the

application of TXOP to the AWRR, packet loss for

high priority data can be reduced by 7%, 5.9% for

medium priority data and 6.1% for low priority data.

The AWRR strategy does not have an internal

collision mechanism in the nodes and also the nodes

only contend for the medium after it is decided

which device will access the medium. Little, if any

improvement with EDCA is observed with EDCA in

WMNs due to the starvation and internal contention

mechanism present. However, TXOP application to

AWRR has shown significant packet loss and end-

to-end delay reduction for all data priority classes.

With AWRR a high increase in end-to-end delay

for high and medium priority packet is observed

compared to EDCA as AWRR gives higher chances

for packets from other priority data classes to be

transmitted on the medium. In a multi-hop scenario,

this results in the end-to-end delay increase for high

and medium priority data only. The TXOP period

helps lower this delay by a significant amount. A

delay reduction of 23ms for high priority, 9ms for

medium and 7ms for low priority data are observed.

This paper has shown that the performance of

AWRR can be further improved and optimised by

the use of TXOP limit values. In EDCA, back-off is

performed concurrently between the parallel queues,

while with AWRR, it is performed after the packet is

selected for scheduling which shows more positive

results with the use of TXOP.

Simulation results show an improvement in

performance in terms of reliability, end-to-end delay

and fairness. The proposed strategy, AWRR with

TXOP is a promising technique for implementation

in smart rural applications such as smart grid, smart

buildings, smart health and smart agriculture as a

low cost option to build and extent networks as

observed through simulation results.

The Impact of Transmission Opportunity (TXOP) on the Performance of Priority based Contention based Scheduling Strategies in

Multi-hop Mesh Networks

119

ACKNOWLEDGEMENTS

The authors will like to thank the reviewers for their

comments. This research was supported by the

University of Botswana and the South African

National Research Foundation (NRF) under the

THRIP project TP13081327740.

REFERENCES

Akyildiz, I.F., Wang, X. & Wang, W., 2005. Wireless

mesh networks: a survey. Computer Networks, 47(4),

pp.445–487.

Chen, Y., 2011. High Performance Distributed

Coordination Function with QoS Support in IEEE

802.11e Networks. Australasian Telecommunication

Networks and Applications Conference (ATNAC).

Deng, J. & Han, Y.S., 2009. Fairness Index Based on

Variational Distance. Global Telecommunications

Conference, pp.1 –6.

Denko, M. & Obaidat, M., 2009. Fairness and Throughput

Optimization in Wireless Mesh Networks. Electronics,

Circuits, and Systems, ICECS IEEE International

Conference, pp.824–827.

Farn, J. & Chang, M., 2005. Proportional Fairness for QoS

Enhancement in IEEE 802.11e WLANS. International

Conference on Local Computer Networks, (1), pp.4–5.

Gao, D., Cai, J. & Ngi, K., 2005. Admission control in

IEEE 802.11e wireless LANs. IEEE Network, 19(4),

pp.6–13.

Guesmi, H. & Maaloul, S., 2013. A Cross-Layer Qos

Based Scheduling Algorithm WRR Design in Wimax

Base Stations. American Journal of Electrical and

Electronic Engineering, 1(1), pp.1–9.

Hammond, A. & Paul, J., 2006. A New Model for Rural

Connectivity. World Resouces Institure, (May).

Hu, J. et al., 2012. Comprehensive QoS Analysis of

Enhanced Distributed Channel Access in Wireless

Local Area Networks. Inf. Sci., 214, pp.20–34.

iDirect, 2009. Eight Essentials to Implementing Backhaul

over Satellite for Mobile Operators. White Paper.

Inan, I., Keceli, F. & Ayanoglu, E., 2007. Analysis of the

802.11e Enhanced Distributed Channel Access

Function. , (0434928), pp.1–36.

Kaveh Pahlavan, P.K., 2002. Principles of Wireless

Networks,

Kuppa, S. & Prakash, R., 2004. Service differentiation

mechanisms for IEEE 802.11 based wireless networks.

Wireless Communications and Networking

Conference, 4, pp.796–801.

Lee, J.F., Liao, W. & Chen, M.C., 2005. A MAC-Layer

Differentiated Service Model in IEEE 802.11e

WLANs. Global Telecommunications Conference, 6,

pp.3290–3294.

Maamar, S. et al., 2011. Contention Window

Optimization: an enhancement to IEEE 802.11 DCF to

improve Quality of Service. International Journal of

Digital Information and Wireless Communications

(IJDIWC), 1(1), pp.273–283.

Min, G., Hu, J. & Woodward, M.E., 2008. An Analytical

Model of the TXOP Scheme with Heterogeneous

Classes of Stations. , pp.1–5.

Pan, S., Wu, J. & You, M., 2009. Collision-Aware

Adaption of Contention Window in 802.11E Wirless

LAN. International Journal of Computer Networks &

Communications (IJCNC), 1(2), pp.12–24.

Pathak, P.H. & Dutta, R., 2011. A Survey of Network

Design Problems and Joint Design Approaches in

Wireless Mesh Networks. IEEE Communications

Surveys & Tutorials, 13(3), pp.396–428.

Periyasamy, P; Karthikeyan, E., 2014. Comparative

Performance Analysis of AODV and AODV-MIMC

Routing Protocols for Mobile Ad hoc Networks.

International Jounal Computer Network and

Information Security, 6(6), pp.54–60.

Poonguzhali, A., 2014. Performance Evaluation of IEEE

802. 11e MAC Layer Using Cell Processor.

International Journal of Sceintific and Technology

Research, 3(1), pp.255–261.

Reddy, T.B., John, J.P. & Murthy, C.S.R., 2007. Providing

MAC QoS for multimedia traffic in 802. 11e based

multi-hop ad hoc wireless networks. Computer

Networks, Elsevier, 51, pp.153–176.

Sheikh, S.M., Wolhuter, R. & Engelbrecht, H.A., 2015.

An Adaptive Congestion Control and Fairness

Scheduling Strategy for Wireless Mesh Networks. In

IEEE Symposium on Computational Intelligence for

Communication Systems and Networks (CIComms15).

Sheikh, S.M., Wolhuter, R. & Rooyen, G.J. Van, 2015. A

Cross-Layer Adaptive Weighted Round Robin

Scheduling Strategy for Wireless Mesh Networks. In

Southern Africa Telecommunication Networks and

Applications Conference (SATNAC). pp. 323–328.

Suzuki, T., Noguchi, A. & Tasaka, S., 2006. Effect of

TXOP – Bursting and Transmission Error on

Application – Level and User – Level QoS in Audio –

Video Transmission with IEEE 802. 11e EDCA.

Telenor, R., Unik, D. & Østerbø, O.N., 2005.

Differentiation of Downlink 802.11e Traffic in the

Virtual Collision Handler. IEEE Conference on Local

Computer Networks (LCN’05).

Xylomenos, G. & Polyzos, G.C., 1999. TCP and UDP

Performance over a Wireless LAN. IEEE INFOCOM,

(March), pp.439–446.

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

120