A Random Priority based Scheduling Strategy for Wireless Sensor

Networks using Contiki

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: Contiki, MAC, IEEE802.15.4, Priority Scheduling, Sensor Networks.

Abstract: In recent years, wireless sensor networks (WSNs) have experienced a number of implementations in various

implementations which include smart home networks, smart grids, smart medical monitoring, telemetry

networks and many more. The Contiki operating system for wireless sensor networks which utilises carrier

sense multiple access with collision avoidance (CSMA/CA) does not provide differentiated services to data

of different priorities and treats all data with equal priority. Many sensor nodes in a network are responsible

not only for sending their sensed data, but also forwarding data from other nodes to the destination. In this

paper we propose a novel priority data differentiation medium access control (MAC) strategy to provide

differentiated services called Random Weighted Scheduling (RWS). The strategy was implemented and

tested on the FIT IoT-lab testbed. The strategy shows a reduction in packet loss compared to the default

CSMA/CA scheduling strategy in IEEE802.15.4 WSNs when carrying data of different priority levels.

1 INTRODUCTION

Wireless Sensor Networking (WSN) is one of the

main driving forces of the Internet of Things (IoT).

WSN have been deployed in a number of different

environments which include smart home networks,

smart health, smart transport, smart educations and

other IoT applications. All these networks carry

heterogeneous data with different levels of priority

(Glaropoulos et al., 2014). A WSN typically consists

of a large number of low cost and low power,

multifunctional sensor nodes. Sensor nodes are

normally equipment with different types of sensors

depending on the parameter they will be measuring,

different embedded microprocessors, different

operating systems, and different radio transceivers

(Jun Zheng, 2009).

WSN embedded operating systems include

among others TinyOS, Contiki, MANTIS, T-Kernal,

LiteOS and Nano-RK. The Contiki operating system

is one of the most popular operating systems for

embedded systems and IoT applications

(Glaropoulos et al., 2014). Contiki utilises the

CSMA/CA scheduling strategy. The standard

CSMA/CA mechanism does not provide any data

differentiation services to improve the quality of

service for time critical events such as alarms that

have a higher priority than normal data in a network

(Koubaa et al., 2006). CSMA/CA treats all data with

equal priority in a first in first out (FIFO) manner.

In this paper we propose a novel scheduling

strategy that has been developed under the Contiki

operating system and implemented and tested on the

FIT IoT-lab testbed. Our proposed scheme has

multiple queues for data of different priority and

smaller values of back-off exponent (BE) and

contention window (CW) are assigned to higher

priority data queues to gain access to the channel

faster than the lower priority queues. The data from

the different queues gain access to the channel by

randomly selecting a queue for transmission based

on weights assigned to the different queues.

The rest of this paper is organised as follows.

Section 2 presents the related work. Section 3

presents an overview of CSMA/CA in the

IEEE802.15.4 standard. In section 4, we present the

proposed RWS scheduling strategy. Section 5

presents an overview of the Contiki operating

system. Section 6 gives a brief overview of the

Testbed implementation. Section 7 presents the

results and in section 8 we conclude the paper.

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

A Random Priority based Scheduling Strategy for Wireless Sensor Networks using Contiki.

DOI: 10.5220/0005949301210128

In Proceedings of the 13th International Joint Conference on e-Business and Telecommunications (ICETE 2016) - Volume 6: WINSYS, pages 121-128

ISBN: 978-989-758-196-0

121

2 RELATED WORK

A large amount of work has been carried out to

optimize energy usage in WSN communications.

Limited work has been done to develop priority

based scheduling strategies in WSNs. A priority

scheduling scheme is proposed in (Sun and Xu,

2010) which is based on queue management and

MAC layer back off. When a packet arrives at the

node, it gets placed at the end of the queue in a FIFO

queue that does not differentiate the priorities of

packets. DRAG (Sun and Xu, 2010) is a priority-

based queue management policy to provide priority

guarantee. The packet gets placed in the queue in an

appropriate place relative to the previously sorted

packets instead of placing the packet at the end of

the queue. Furthermore, the strategy selects a high

priority packet to send.

Other MAC layer priority based scheduling

strategies have also been proposed in literature such

as Q-MAC and PRIMA (Barua et al., 2014). With

Q-MAC, a queuing model with multiple queues for

different priority levels packets is proposed. The

strategy tries to minimize energy consumption and

provides QoS for intra-node and inter-node

scheduling. The intra-node scheduling places data

into the different priority queues, while the inter-

node scheduling provides channel access to reduce

energy consumption by reducing collisions. Five bits

of information are added to the existing packets of

which 2 are for identification of the packet type and

the other 3 are for sensing data. Packets that have

travelled more hops have a higher priority. In Q-

MAC, the queue architecture consists of five queues

with one specified as an instant queue. PRIMA is

also an energy efficient MAC protocol which

minimizes the idle listening periods by making

nodes that have no data to send to go to a sleeping

state. PRIMA also employs multiple queues where

data is classified and placed in the respective queues.

Queues with higher priority are given first access to

the channel compared to the low priority data.

To provide service differentiation to rate

sensitive applications, (Na, 2011) proposes a Multi-

rate Service Differentiation (MSD) model. This

model is implemented with two components namely

a Virtual Medium Access Control (VMAC) and the

Adaptive Back-off Window Control (ABWC). The

VMAC is the priority queue mechanism that adapts

its back-off based on the conditions of the network.

With VMAC, it is possible that more than one data

packet collided with each other if they finish the

back-off period at the same time. A Virtual Collision

Avoidance Control (VCAC) is designed to address

this situation which adjusts the back-off times of

data frames. In (Koubaa et al., 2006), a mechanism

that tunes the existing parameters of CSMA/CA for

data of different priority is proposed. These include

BE and CW values. This strategy differentiates

between data traffic and command traffic in a

network. Command traffic are given higher priority

by assigning smaller BE and CW values.

The above priority strategies are implemented in

sensor networks working on the IEEE802.15.4

standard. In the IEEE 802.11e standard, a contention

based strategy called enhanced distributed channel

access (EDCA) is used to provide differentiated

services for data of different priority levels. With the

default EDCA strategy, an unfairness problem is

known to exist between higher and lower priority

data classes as higher priority data can starve lower

priority data (Choi et al., 2008; Kuppa and Prakash,

2004; Tseng et al., 2007). EDCA introduces the

concept of access categories (AC) for data types and

consists of four ACs. Each AC has specific

parameters associated to its priority class such that

the higher probabilities ACs have a higher

probability of medium access (Poonguzhali, 2014).

Q-MAC is a complex strategy that introduces

extra overhead in the network by the introduction of

5 extra bits added to every message designed for

energy conservations. There are many applications

such as smart home networks, smart grids, smart

medical monitoring and telemetry networks where

some nodes can be designed to act as backbone

nodes and relay information from these clusters to

the destination. In our proposed scheme, we use the

same concept as Q-MAC by having multiple queues

for data of different priority and we assign smaller

values of BE and CW for higher priority data.

However, the design does is not concerned with

energy conservation and scheduling to reduce

energy wastage and therefore, we do not compare its

performance to Q-MAC.

3 OVERVIEW OF CSMA/CA IN

THE IEEE 802.15.4 STANDARD

The popularity and features of machine-to-machine

communications and the Internet of things (IoT),

have resulted in a wide areas of research leading to

development a low-power, low-rate, and low-cost

wireless system. The IEEE 802.15.4 standard has

become a standard for low rate wireless personal

area network (LR-WPAN) communications (Hwang

and Nam, 2014). The IEEE 802.15.4 standard which

operates at the link and physical layers is designed

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

122

for simple, low data rate, low-power and low-cost

wireless communication with wireless personal area

networks (WPANs). It is implemented in wireless

sensor networks. The unlicensed Industrial,

Scientific and Medical (ISM) band that operates

worldwide with this technology is the 2.4 GHz ISM

band (Petrova et al., 2006). In this band of 2.4 GHz,

the ISM offers 16 channels with a data rate of 250

kbps (Collotta et al., 2013). Wireless data exchange

is done through the direct sequence spread spectrum

(DSSS) modulation scheme (Petrova et al., 2006). In

our study we use a radio model that also uses the 2.4

GHz ISM band.

According to this standard, a node can optionally

operate in beacon-enabled mode or non beacon

enabled mode (Collotta et al., 2013). In this section,

we present a brief overview of the beacon enabled

mode which is based on the slotted mode; and the

non-beacon enabled mode CSMA/CA mechanism of

the IEEE 802.15.4 standard which is based on the

un-slotted mode. Our work is based on the un-

slotted, non-beacon enabled mode CSMA/CA

mechanism to access the channel and transmit data.

3.1 Beacon Enabled Mode

For the slotted mode, the slots are aligned with the

beacon frames sent periodically by the Personal

Area Network (PAN) coordinator. With the un-

slotted mode, there are no beacon frames (Kim et al.,

2007). The principle of operation of this standard

depends on beacon messages in the form of

superframes regularly sent from a PAN coordinator.

The MAC superframe structure is shown in fig 1.

The time between the beacons is split into 16 slots.

The superframe consists of an active period with the

Contention Access Period (CAP) and CFP

(Contention Free Period), and an inactive period.

The CFP consists of Guaranteed Time Slots (GTS)

which are allocated to support QoS such as real-time

applications (Youn et al., 2007). In the CFP region,

nodes can obtain access to the medium without

collisions. In the inactive period the radio interface

can be put in a low energy consumption status in

order to improve energy savings (Collotta et al.,

2013). GTS are provided by the PAN coordinator for

nodes that need to transmit data within a certain

time. Nodes access the CAP using CSMA/CA. In

this mechanism, a node that wants to send data first

senses the medium after a random number of back-

off periods. If the medium is free the data is

transmitted, otherwise a back-off is performed.

There are seven GTS slots that can be

accommodated in one frame. There are limitations

Figure 1: The MAC super frame structure in the IEEE

802.15.4 standard.

Figure 2: The slotted CSMA/CA mechanism in IEEE

802.15.4.

Figure 3: Flow chart for the non-beacon enabled unslottted

mode for the CSMA/CA mechanism with ACK in the

IEEE 802.15.4 standard.

with GTS as it can only support a limited number of

nodes and does not provide any method to support

QoS in the CSMA/CA mode (Youn et al., 2007).

A Random Priority based Scheduling Strategy for Wireless Sensor Networks using Contiki

123

An overview of CSMA/CA in slotted mode is

presented in fig 2. NB denotes the number of times

that the algorithm is required to back-off due to the

medium being busy during channel assessment. CW

is the contention window which is the number of

back-off periods that need to be clear of channel

activity before the packet transmission can be

started. BE is the back-off exponent which is the

number of back-off periods that a device should wait

before attempting to assess the channel.

When a packet arrives, NB, BE and CW are

initialized (ie. NB = 0, CW = CWinit = 2, BE = 2 or

BE = min(2,macMinBE) where macMinBE is the

default minimum BE value). After initializing of the

variables, the back-off period is started which is

chosen by a random number generated in the range

of [0, 2

− 1]. When this back-off has expired, the

algorithm then performs one Clear Channel

Assessment (CCA) to verify if the channel is busy or

free. If the channel is found to be busy, the CW is

again initialized to CWinit = 2, the NB and BE

variables are incremented by one. If the channel is

found to be free (idle), the CW is decremented by

one. The CCA process is than repeated until the CW

value becomes 0. After this the data is transmitted.

This mechanism ensures that at least two CCA

operations are performed to prevent potential

collisions (Youn et al., 2007). If the channel is busy,

both values of NB and BE are increased by one. BE

cannot exceed the set macMaxBE having the default

value 5 and CW is reset to 2. If NB becomes greater

than the set maximum back-offs allowed, the

algorithm terminates with a channel access failure

status. This failure will be reported to the higher

protocol layers, which decide whether or not to

attempt the transmission as a new packet again (Kim

et al., 2007).

3.2 Non-Beacon Enabled Mode

Just like in the beacon enable mode, when a packet

arrives, the number of back-offs (NB) and the back-

off exponent (BE) are initialized. After this

initialization of the variables, the back-off period is

started which is chosen by a random number

generated in the range of [0, 2

− 1]. Initially, BE is

initialized to BEmin which is 3 by default. BEmax is

5 by default. When this back-off has expired, the

algorithm then performs one CCA to verify if the

channel is busy or free. If the channel is found to be

busy, the NB and BE variables are incremented by

one. The procedure is repeated until NB is less than

the set maximum allowed transmissions. After the

maximum Transmissions allowed set + 1

unsuccessful attempts to access the channel, the

packet is dropped. If the channel is found to be free

(idle), a transmission takes place.

In the IEEE802.15.4 standard, the

acknowledgement (ACK) mode to transmitted

packets is optional unlike in the IEEE 802.11

standard. It is an optional feature as it can increase

network overhead and have an effect on the

achievable throughput of the network. If ACK mode

is enabled, for any transmission that does not receive

an acknowledgment, the NB and BE values are

increased. If NB becomes greater than the set Max

Transmissions allowed, the algorithm terminates

with a channel access failure status (Kim et al.,

2007). Fig 3 presents the flowchart for the operation

of CSMA/CA in non-beaconed un-slotted mode.

4 THE PROPOSED RANDOM

WEIGHTED SCHEDULING

(RWS) STRATEGY

This section presents an overview of the proposed

strategy. In each node three queues are created.

These are for high, medium and low priority data. In

the packet, a data field is created of 2 bits which

carries information on the priority of the packet.

Using this information, data is placed in either one

of the 3 queues depending on the priority set in the

packet header. The priority field in the packet is

shown in fig 4. When a node has data to transmit

and more than one queue has data, a selection

process is followed. If only one queue has data, then

the packet from that queue is selected for

transmission without the need to follow a selection

process. However, the BE, CW and NB processes

are carried out after the selection process as is in the

un-slotted non-beaconed based CSMA/CA. With our

strategy we assign smaller values of BE and CW for

higher priority data to allow the higher priority data

to gain access to the channel faster than the lower

priority data. The queue selection followed when

two or more queues have data is as follows:

1. Probability weights are initially assigned to each

data priority queue.

2. The strategy determines the size of the individual

queues. If all of the queues have data, the

original assigned weights in stage 1 are used. If

all the queues do not have data, then the weight

of the queues with data are added and then the

weights of the queues with data are normalised

and assigned new weights. The queues with no

data are assigned a weight of zero i.e. 0.

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

124

Figure 4: Integration of priority information in the packet.

Figure 5: Queue Selection process in RWS.

Figure 6: Proposed RWS strategy mechanisms.

3. The range values for each of the priority data

classes are assigned over a scale with the high

priority data being first on the scale, then

medium priority data and lastly low priority data.

4. A random number is then generated in the range

of 0 to the maximum scale value. A packet is

chosen for transmission from a queue from

which the number generated falls in its range as

shown in fig 5.

5. After the packet is selected, the BE, CW and NB

processes are followed as stated earlier. The

complete scheduling strategy is shown in fig 6.

5 CONTIKI

The scheduling scheme proposed in this paper was

implemented in Contiki as Contiki does not have a

priority based scheduling mechanism for data of

different priority. In this section we give a brief

overview of this Contiki operating system which is

an open source operating system.

Contiki is implemented in the C language

developed at the Swedish Institute of Computer

Science (SICS) (Dunkels, 2004); (Networks, 2011).

Contiki has an event-driven kernel and follows a

linear programming style which was used for the

programming in this work. The Contiki protocol

stack is designed for resource-constrained devices

with constraints on memory and processing power

Figure 7: Contiki Layer Model.

(Colitti et al., 2009). It supports IPv6, RPL routing

protocol for low-power and lossy networks, Rime

and the Constrained Application Protocol (CoAP),

making it suitable to develop IoT applications

(Glaropoulos et al., 2014). Compared to many other

closed source firmware operating systems

implemented in hardware, Contiki is open source as

stated earlier. We therefore, used Contiki in our test

bed implementation as it allows us to use, modify

and make additions to this operation system.

The Contiki OS provides two communication

stacks namely uIP and Rime. uIP is a TCP/IP stack

that makes it possible for Contiki to communicate

over the Internet. Rime is a lightweight

communication stack designed for low-power radio

communication. Rime is a custom lightweight

networking stack. It provides primitives for single-

hop and multi-hop (mesh) communication

(Networks, 2011). In our study, the Rime

communication stack for multi-hop communication

was used as the other layers are less detailed. Fig 7

presents the communication protocol stack used in

our study. Our scheduling strategy is implemented at

the MAC layer as an enhancement to CSMA/CA.

6 TESTBED IMPLEMENTATION

The FIT IoT-Lab testbed was used to implement and

test the RWS scheduling strategy. The FIT IoT-Lab

is a very large scale open WSN test bed at INRIA,

France. This test bed allows for testing of scalable

protocols and applications on this large scale test

bed (Rosiers et al., 2011); (Inria, 2016).The

implementation and testing of the scheduling

strategy was done on nodes in Lille and Grenoble.

The strategy was implemented on M3 nodes which

has a 32-bit ARM cortex micro-controller, high

performance and uses a 2.4GHz radio interface.

A Random Priority based Scheduling Strategy for Wireless Sensor Networks using Contiki

125

Table 1: RWS parameters.

Traffic Type BE CW

min

max

High Priority Data 4 7 15

Medium Priority Data 5 15 31

Low Priority Data 10 31 1023

Table 2: Data Transmission Test Cases.

High Priority Data

Medium Priority

Data

Low Priority

Data

Case 1 60 packets/sec 60 packets/sec 60 packets/sec

Case 2 60 packets/sec 120 packets/sec 120 packets/sec

Case 3 120 packets/sec 60 packets/sec 120 packets/sec

The scheduling strategy is mainly developed for

backhaul nodes that will carry data of different

priority in a multi-hop fashion until it reaches its

destination. The M3 nodes were chosen over other

available test bed nodes such as the WSN430 and

A8 nodes as they have much higher processing

power, which is needed in backbone WMN nodes.

The nodes that were chosen as one hop away were

spaced 4.8m apart. The mesh network was setup so

that communication with the receiver takes place in

multiple hop fashion by limiting the transmission

range of the nodes. There are two ways of limiting

the transmission range. These are being either

decreasing the transmission power, or by setting a

minimal energy level for the packet reception. The

range of the nodes was limited so that it can only

communicate with its 1 hop neighbour.

The default CSMA/CA scheduling strategy in

Contiki works in a FIFO fashion and does not

differentiated packets of different priority. Our RWS

strategy was developed on top of this by introducing

3 queues in the nodes, one for each priority level. An

application was written that generates packets of

100byte of different priority levels at the

transmission rates of the different test cases. Packets

with the fields as shown in fig 5 were created. The

application at each node records the number of

packets sent and number of packets received. Before

implementing the scheduling strategy and

application to generate packets of different priority

level, the codes written in C were tested in the Cooja

simulator on Tmote Sky nodes. It was compiled for

the actual test bed nodes and then implemented on

the test bed.

Since the strategy is intended to be used in

backhaul networks receiving a high number of

packets from the different network domains, the

parameters for BE, NB and CW were adjusted to

match those of the EDCA strategy in the

IEEE802.11 standard. The modifications therefore

made to the CSMA/CA in the IEEE802.15.4 were as

follows:

1. The acknowledgment mechanism was activated

to receive acknowledgment messages for any

successful transmission as in the IEEE 802.11

CSMA/CA; 2. The maximum transmission

allowed value was set to 7 as is the case with

IEEE802.11g; 3. The values of BEmin and

BEmax were changed such that the CW size will

be the same as in the EDCA based on the

CWmin and CWmax values. Table 1 presents the

default parameters used for high, medium and

low priority data to match those used in EDCA.

In the IEE 802.11 standard, the CW length is

basically the number of back-off countdown

periods that need to sense the channel idle before

a transmission attempt can be made while in the

un-slotted IEEE802.15.4, it is the duration that

the back-off waits before performing CCA.

Channel activity is only performed at the CCA

stage; 4. For Distributed Coordination Function

(DCF), BE is set to 10 as BE of 10 equals a CW

size of 1023. The retransmission limit was set to

The proposed RWS strategy was tested against

CSMA/CA in many test cases. Three test cases are

presented in this paper as shown in table 2. The

packets are transmitted at different rates of high and

low combinations for number of hops from 1 to 7

hops. Each intermediate hop forwards data as well as

transmits its own data generated at the rates stated in

table 2. The chain topology was used for the

implementations.

7 RESULTS

The packet loss results for the different test cases

with the default CSMA/CA and the proposed

scheduling strategy are shown in figs 8 to 10. In the

graphs we call CSMA/CA as DCF as it operates like

the DCF in the IEEE802.11 standard, with all data of

any priority level treated in a single queue, FIFO

manner. The results are those obtained from the real

test bed implementation as such the conditions of the

channel can change over time depending on the

environment. The performance of CSMA/CA

depends also on the value chosen for the back-off

which is randomly selected. For any two test bed

tests carried out, the exact conditions might not be

the same as the number generated might be different

which has an effect when the packets are transmitted

to the next hop as well as the link conditions. The

proposed scheme also largely depends on a random

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

126

Figure 8: Packet loss in test case 1.

Figure 9: Packet loss in test case 2.

Figure 10: Packet loss in test case 3.

number generated to choose which queue must

transmit its data. To keep the conditions of the

channel the same, both tests for each hop number

were run immediately one after the other for

CSMA/CA and RWS for the same hop number to

make the comparison approximately the same.

For all the three test cases, the tests were run for

five minutes each. Less packet loss can be observed

for RWS over CSMA/CA in nearly all the test cases

as the number of hops increases. The performance at

2 or 3 hops is approximately the same as for the

original CSMA/CA technique. The performance

improvement is mainly observed at hops more than

3. There is also more packet loss observed at higher

loads (case 3), compared to lower loads (case 1).

When the data loads are the same, a higher packet

loss for the lower priority data is observed. All these

test bed results presented in this paper conform to

the 95% confidence levels.

8 CONCLUSIONS

A novel scheduling strategy for networks carrying

data of different priority has been developed and

implemented in the Contiki operating system.

Contiki is an open source operating system and

therefore, we modified the existing codes to

implement our strategy. The Rime protocol

communication stack was used as the other layers

are light weight and this helps to ascertain the

performance of the proposed scheme.

The proposed strategy has shown a reduction in

packet loss as the number of hops is increased for

most of the test cases implemented over the FIT IoT-

lab test bed. The assessment of the performance of

the strategy by means of a live test bed is more

accurate, as opposed to testing by simulation only.

This is clearly beneficial in terms of confidence in

A Random Priority based Scheduling Strategy for Wireless Sensor Networks using Contiki

127

the success of any future implementations. This

work is important in view of the rapid IoT and smart

application implementations.

ACKNOWLEDGEMENTS

The authors will like to thank the reviewers for their

comments; INRIA (France) and Nathalie Mitton for

letting us use the FIT IoT testbed; and Viktor Toldov

for assisting with the use of the test bed. This

research was supported by the University of

Botswana and the South African National Research

Foundation (NRF) under the THRIP project

TP13081327740.

REFERENCES

Barua, S., Afroz, F., Islam, S. S., Ahmed, A. U., Ghosal,

P., Sandrasegaran, K. (2014). Comparative study on

Priority Based QoS Aware MAC Protocols for WSN.

International Journal of Wireless & Mobile Networks

(IJWMN), 6(5), 175–181.

Choi, J., Yoo, J., Kim, C. (2008). A Distributed Fair

Scheduling Scheme with a New Analysis Model in

IEEE 802.11 Wireless LANs. IEEE Transactions on

Vehicular Technology, 57(5), 3083–3093.

Colitti, W., Steenhaut, K., Lemmens, B., Borms, J. (2009).

Simulation Tool for Wireless Sensor Network

Constellations in Space.

Collotta, M., Scatà, G., Pau, G. (2013). A Priority-Based

CSMA/CA Mechanism to Support Deadline-Aware

Scheduling in Home Automation Applications Using

IEEE 802.15.4. International Journal of Distributed

Sensor Networks, 2013.

Dunkels, A. (2004). Contiki - a Lightweight and Flexible

Operating System for Tiny Networked Sensors. In

Proceedings of the First IEEE Workshop on

Embedded Networked Sensors, Tampa, Florida, USA.

Glaropoulos, I., Vukadinovic, V., Mangold, S. (2014).

Contiki80211: An IEEE 802.11 Radio Link Layer for

the Contiki OS. IEEE International Conference on

High Performance Computing and Communications

(HPCC), 621–624.

Hwang, K., Nam, S. (2014). Analysis and Enhancement of

IEEE 802.15.4e DSME Beacon Scheduling Model.

Journal of Applied Mathematics, 2014.

Inria. (2016). FIT IoT-Lab. Retrieved from https://

www.iot-lab.info/hardware/

Jun Zheng, A. J. (2009). Wireless Sensor Networks, A

Networking Perspective. John Wiley & Sons.

Kim, E., Kim, M., Youm, S., Choi, S., Kang, C. (2007).

Priority-based service differentiation scheme for IEEE

802.15.4 sensor networks. International Journal of

AEU of Electronics and Communications, 61, 69–81.

Koubaa, A., Alves, M., Nefzi, B., Song, Y.-Q. (2006).

Improving the IEEE 802.15. 4 slotted CSMA/CA

MAC for time-critical events in wireless sensor

networks. Workshop on Real Time Networks RTN, 1–

Kuppa, S., Prakash, R. (2004). Service differentiation

mechanisms for IEEE 802.11 based wireless networks.

Wireless Communications and Networking

Conference, 4, 796–801.

Na, C. W. (2011). IEEE 802.15.4 Wireless Sensor

Networks: GTS Scheulding and Service

Differentiation.

Networks, S. (2011). Comparison of Simulators for

Wireless Sensor Networks.

Petrova, M., Riihij, J., Petri, M. (2006). Performance

Study of IEEE 802. 15. 4 Using Measurements and

Simulations. IEEE Wireless Communications and

Networking Conference (WCNC), 1, 487–492.

Poonguzhali, A. (2014). Performance Evaluation of IEEE

802. 11e MAC Layer Using Cell Processor.

International Journal of Sceintific and Technology

Research, 3(1), 255–261.

Rosiers, C. B., Chelius, G., Ducrocq, T., Fleury, E., &

Vanda, J. (2011). SensLAB : a Very Large Scale Open

Wireless Sensor Network Testbed, 2011.

Sun, G., & Xu, B. (2010). Dynamic Routing Algorithm for

Priority Guarantee in Low Duty-Cycled Wireless.

Springer-Verlag Berlin Heidelber LNCS, 146–156.

Tseng, K.-N., Wang, K., Shih, H.-C. (2007). Enhanced

Fair Scheduling for IEEE 802.11e Wireless LANs.

Journal of Information Science and Engineering,

1721, 1707–1721.

Youn, M., Lee, J., Kim, Y. (2007). IEEE 802.15.4 based

QoS support Slotted CSMA / CA MAC protocol. In

International Conference on Sensor Technologies and

Applications (pp. 113–117).

WINSYS 2016 - International Conference on Wireless Networks and Mobile Systems

128