An Adaptive Acknowledgement On-demand Protocol for Wireless Sensor

Networks

Cung Lian Sang

, Marc Hesse

, Sebastian Zehe

, Michael Adams

Timm H

ormann

and Ulrich R

uckert

Cognitronics and Sensor Systems Group, CITEC, Bielefeld University, Bielefeld, Germany

Ambient Intelligence Group, CITEC, Bielefeld University, Bielefeld, Germany

Keywords:

Adaptive Protocol, WSN, ACK, No-ACK, Acknowledgement On-demand, TDMA, Schedule-based, Wireless

Sensor Network, CC1101.

Abstract:

The concept of packet acknowledgement in wireless communication networks is crucial for reliable data trans-

mission. However, reliability comes with the cost of an increased duty cycle of the network. This is due to

the additional acknowledgement time for every single data packet sent. Therefore, energy consumption and

latency of all sensor nodes is increased whilst the overall throughput in the network decreases. This paper

contributes an adaptive acknowledgement on-demand protocol for wireless sensor networks with star network

topology. The goal is to tackle the trade-off between energy efﬁciency and reliable data transmission. The pro-

posed protocol is able to detect network congestion in real time by constantly monitoring the overall packet

delivery ratio for each sensor node. In case the packet delivery ratio of any sensor nodes in the network is

dropped signiﬁcantly (e.g. due to environmental changes), the protocol switches automatically to a more reli-

able data transmission mode utilizing acknowledgements concerning the affected sensor nodes. Our proposed

method is tested and evaluated based on a speciﬁc hardware implementation and the corresponding results are

discussed in this paper.

1 INTRODUCTION

In wireless sensor network (WSN) applications, the

concept of packet acknowledgement (ACK) is often

crucial because it is the best way for the transmitter to

know whether the transmitted packets were received

successfully. If the transmitter does not receive the

ACK signal, it is concluded that the packet was not

received, thus the same data needs to be retransmit-

ted. Depending on the type of protocol used, the re-

transmission could be done in the next active time slot

(scheduled based topologies) or it could be done until

a certain time-out is reached (IEEE, 2011). However,

it is obvious that those approaches increase the over-

head of the data frame format and require a certain

delay in order to receive the ACK signal. In turn, this

delay can have a negative effect on long-term energy

efﬁciency of low power WSNs. This is because en-

ergy efﬁciency is strongly correlated with the sensor

node’s duty cycle in WSNs. Thereby, the duty cy-

cle refers to the cyclic ratio of active time (transmit-

ting and receiving) vs. sleep time in schedule-based

topologies.

Reducing ACK time is also directly related to the

throughput in WSNs. Since the throughput is lim-

ited, not only by the transmission time of the pack-

ets, but also by the transmission time of ACK sig-

nals (Takamori and Yamao, 2015). At the same time,

it is also directly related to the energy efﬁciency of

WSNs. The reason behind is that the shorter the data

transmission time, the longer the sensor nodes can

stay in sleep mode. This allows extended operating

times in battery-operated scenarios. This issue is even

more relevant if the packet size of the transmitted data

is comparable to the packet size of the ACK signal,

which is very often the case in WSNs (Takamori and

Yamao, 2015).

The key contribution of this paper is the devel-

opment of an adaptive acknowledgement on-demand

protocol for WSNs, which is able to switch from non-

acknowledgement (No-ACK) to acknowledgement

(ACK) mode and vice versa. It is using time divi-

sion multiple access (TDMA) channel access method.

The switch over between ACK and No-ACK- depends

on the condition of the network congestion. With

the proposed Acknowledgement On-Demand Proto-

174

Lian Sang C., Hesse M., Zehe S., Adams M., HÃ˝urmann T. and RÃijckert U.

An Adaptive Acknowledgement On-demand Protocol for Wireless Sensor Networks.

DOI: 10.5220/0006208501740181

In Proceedings of the 6th International Conference on Sensor Networks (SENSORNETS 2017), pages 174-181

ISBN: 421065/17

col, we aim towards improving the overall energy

efﬁciency, latency, and throughput in WSNs. Fur-

thermore, we describe how the reliability of data

transmission in star network topologies with TDMA

scheme can be achieved effectively without using

ACK. The proposed protocol was tested indoors with

a speciﬁc hardware implementation simulating a typ-

ical smart home environment.

This paper is organized as follows: In section 2,

the related work of the proposed concept is presented,

followed by the system model of the proposed adap-

tive protocol in Section 3. The results of experimen-

tal evaluation and summary are presented in Section

4. The current limitations of the proposed protocol is

described in section 5 and ﬁnally, the paper is con-

cluded in section 6.

2 RELATED WORK

We found that very few papers in the literature explore

how to efﬁciently reduce the packet ACK time in sen-

sors networks for reliable data transmission. Con-

cerning this, we brieﬂy discuss previous related work

that supports and highlights our proposed approach in

this section.

In this regard, the group ACK method in a star

topology of sensors network (Takamori and Yamao,

2015) was proposed to reduce the duration of the

ACK time. With this approach, the network perfor-

mance was improved and the bottleneck problem of

the sink node (or central node), where the communi-

cation trafﬁcs are concentrated, was reduced. Also,

the turnaround time of the antenna and the duration

of total ACK time was reduced. This is because in-

stead of sending individual ACK signals for every sin-

gle transmitted data packet, a multiplexed group ACK

signal is sent once in the preceding communication

frame. The variable-period group ACK method is the

best choice in all of the three methods for reducing

the average time delay. Also, both ﬁxed- and variable-

period ACK methods outperform the normal individ-

ual ACK method. With their contribution, they high-

light how the extended time frame due to ACK signal

affects the performance of the whole WSNs. How-

ever, their approach is not based on network conges-

tion and could lead to unnecessarily sent ACK mes-

sages.

Moreover, the comparison of ACK and No-ACK

modes in sensors networks on end-to-end delay and

throughput was analysed by (Al-Sharbaty, 2014). In

their work, ﬁxed and mobile sensor nodes are com-

pared for tree and mesh topologies using the Zig-

bee protocol

. According to the simulation results,

the choice of ACK or No-ACK inﬂuences the per-

formance of the network, which also depends on the

topology type that is used in the above mentioned two

states of sensor nodes. According to the authors’ ﬁnd-

ings, the end-to-end delay of the No-ACK is gener-

ally shorter than ACK in both ﬁxed and mobile sensor

nodes states. In the same way, the overall throughput

of No-ACK is generally higher compared to ACK in

both mentioned states.

Packet loss analysis for the No-ACK mode of

IEEE 802.15.4 MAC (Shu et al., 2007) was presented

with a non-stationary Markov Chain on the beacon

enabled star topology. Moreover, the accuracy of the

model was veriﬁed with the simulation results. It was

concluded that the packet loss rate in No-ACK, in

general, will increase in a network with more nodes

and bigger packet size. This could be overcome with

our approach by taking advantage of prior knowledge

of reliable data transmission in a speciﬁc network sce-

nario.

A dynamic adaptive acknowledgement strategy

was presented by (De Oliveira and Braun, 2005).

They improves the performance of Transmission Con-

trol Protocol (TCP) in multi-hop wireless networks by

dynamically adjusting channel conditions. Another

approach that effectively performs congestion con-

trol in WSNs was presented by (Scheuermann et al.,

2008). Their evaluations are speciﬁcally focused on

the issue of TCP congestion control in the transport

layer of wireless ad-hoc multi-hop networks. How-

ever, the impact of ACK in single hop centralized net-

work scenario was not fully investigated.

In this paper, the presented protocol is based

on TDMA scheme because it is one of the promis-

ing approaches for latency and energy efﬁcient pri-

oritized applications in low power WSNs (Hadded

et al., 2015). For instance, the TDMA-based e-health

WSNs (Gama et al., 2010) and a resource allocation

scheme in wireless body sensor networks (Liu et al.,

2016) were implemented with TDMA to optimize

both energy efﬁciency and quality of service (QoS).

Another example of TDMA can be found for WSNs

operating in noisy environments where it is used to

avoid packet collision (Montiel and C

ardenas, 2014).

Besides, the comparative analysis of a contention

based and TDMA based MAC protocols for WSNs

(Chand and Kakria, 2015) has shown that the TDMA

based protocol outperforms contention based proto-

cols in terms of averaged end-to-end delay, packet de-

livery ratio, and average energy consumption.

By far, all related work mentioned in this section

focuses on the best and appropriate static method of

http://www.zigbee.org/

An Adaptive Acknowledgement On-demand Protocol for Wireless Sensor Networks

175

the protocol for WSNs before the network is actually

deployed in a speciﬁc area. Thus, they are designed

to perform well in a static situation. However, there

are certain cases, in which the trade-off between en-

ergy and latency efﬁciency made is unnecessary and

can be optimized if prior knowledge of the network

congestion is effectively applied.

However, our proposed protocol differs in many

ways from that of the above-mentioned approaches.

Firstly, the decision whether No-ACK or ACK mode

should be used for a speciﬁc sensor node is deter-

mined by the central node itself after the network is

properly deployed in a certain area. Secondly, the

proposed protocol has the ability to adapt to the cur-

rent network congestion and is able to switch from

No-ACK to ACK mode for particular sensor nodes if

necessary in order to maintain the highest energy efﬁ-

ciency. Calculation of packet delivery ratio (PDR) be-

tween transceivers for reliable data transmission with-

out using ACK in TDMA based WSNs is presented

in this paper. To realize energy efﬁcient data trans-

mission, our proposed protocol uses TDMA without

ACK (No-ACK) for the sensor nodes in the uncon-

gested network scenario and TDMA with ACK is

used for the congested network scenario. This dual

approach is chosen in order to increase the reliability

of the data transmission while maximizing the energy

efﬁciency. The implementation details and test results

of our proposed adaptive protocol are given in section

3 and 4.

3 SYSTEM MODEL

3.1 Hardware Design

In order to test our proposed networking concept in

hardware, we implemented exemplary sensor nodes

and a central node using the BRIX

prototyping plat-

Figure 1: Base module and extension modules of BRIX

form

. BRIX

(Zehe et al., 2012) consists of a base

module that can be combined with optional exten-

sion modules (Fig. 1). These extension modules al-

low to add functionalities to match a desired appli-

cation. The BRIX

base module contains an Atmel

ATmega32U4 and an ATmega328P microcontroller,

a Texas Instruments CC1101 Radio Frequency (RF)

transceiver operating in the 868 MHz ISM band, an

Invensense MPU9150 IMU as well as a 450 mA Li-

Poly battery. The compact form factor of BRIX

and

the software framework based on Arduino

allowed

us to rapidly develop, modify and deploy an exem-

plary WSNs.

3.2 Data Frame Format of the Proposed

Protocol

All conducted tests and results presented in this pa-

per are solely obtained by using BRIX

devices. To

reduce the overhead, only a single data format is used

for both the receiver and transmitter in our proposed

protocol (Fig. 2). The protocol is evaluated with star

network topology. The message size of the physical

layer for BRIX

is 29 bytes with a user control data

payload of 20 bytes. The user control frame is com-

posed of the senders address, the frame control sec-

tion, the length of the message sent and the payload.

The frame control section is responsible for control-

ling the message ﬂow in the network. The reason why

we use only a sender address in our proposed imple-

mentation, is that the data transmission in a star net-

work topology is simply between the central node and

the sensor nodes. Thus, it can easily be identiﬁed with

just the sender’s address. The data frame format for

the physical layer has already been deﬁned in the li-

brary of BRIX

Preamble

SFD

MAC Data Payload

(User Control Data Payload)

2 B

1 B

20 B

4 B

Length

FCS

Address

Frame

Ctrl

Length

Data Payload Unit

2 B

PHY Frame Data Format of BRIX

MAC or User Control Frame Data Format of BRIX

2 B

14 B

Figure 2: Data frame format of the proposed algorithm with

minimum overhead.

3.3 Time Slot Allocation in TDMA

In TDMA, each sender is allocated to a certain time

https://www.techfak.uni-bielefeld.de/ags/ami/brix2/

https://www.arduino.cc/

SENSORNETS 2017 - 6th International Conference on Sensor Networks

176

slot. With this mechanism, a single carrier frequency

can be shared among different sensor nodes without

risking packet collisions. Figure 3 describes the time

slot allocation for TDMA scheme implemented in

BRIX

device for testing the proposed protocol. The

time slot for each sensor node in TDMA depends on

the number of bytes that are to be transferred. This

number changes with the packet size and the decision

between ACK and No-Ack mode. It follows that duty

cycle and throughput of a sensor node in TDMA are

directly associated with the active time of that sen-

sor node and the FrameTimeSlot of the network. In

fact, the active time depends on the size of the data

payload and the ACK or No-ACK mode (Fig. 3). In

the same manner, the FrameTimeSlot depends on the

active time of every sensor node in the network with

their corresponding guard time slot and the number of

the sensor nodes in the network.

3.4 Algorithm of the Proposed Adaptive

Protocol for WSNs

The duty cycle is the key factor for energy efﬁ-

ciency in TDMA based WSNs as previously men-

tioned. Therefore, our proposed protocol is evalu-

ated in this manner. This is because the energy con-

sumption of a sensor node in WSNs is determined

by the sensor nodes’ current consumption in transmit,

idle and sleep mode, which is 16.8 mA, 1.7 mA and

100 µA in BRIX

. To maximize the energy efﬁciency

and maintain the reliability of the network, we pro-

posed an adaptive acknowledgement on-demand pro-

tocol for WSNs. The ﬂow chart of our proposed pro-

tocol is depicted in Figure 4 and could be summarized

as follows:

• An initial network congestion test is done by

sending predeﬁned packets without ACK. This is

done after the sensor nodes are actually deployed

in a speciﬁc network area.

• If the packet delivery ratio (PDR) of a sensor

node in the network is higher than the predeﬁned

threshold value, the central node uses the No-

Time (ms)

Sensor

Node 1

Sensor

Node 2

Sensor

Node 3

Sensor

Node N

Tf = Frame Time Slot Ts = Single Time Slot (Node)

Ta = Active Time Tg = Guard Time

Figure 3: TDMA scheme of the proposed protocol.

ACK mode for this particular sensor node. Oth-

erwise, ACKs are used.

• During data transmission using the No-ACK

mode, a continuous network congestion test is car-

ried out. It is done by constantly monitoring the

PDR of each sensor node. This allow to react to

temporary changes in the network. Based on the

PDR soft margin threshold, automatically switch-

ing to the ACK mode is done in case failure of one

or more sensor nodes is detected.

• In order to identify whether switching to the ACK

mode is caused by a temporary or permanent

changes in the network environment, the protocol

is able to switch back to the energy efﬁcient No-

ACK mode once an adjustable countdown timer

runs out.

The assurance of PDR is the central process in

the proposed algorithm. The monitoring process of

the algorithm is mainly based on counting how many

times data from a certain sensor nodes is received by

the central node. The PDR calculation for No-ACK

mode is based on the received packets counter and

the TDMA frame cycle time.

For a reliable and energy efﬁcient data transmis-

sion, the proposed protocol allows a mixed mode

which is ACK and No-ACK modes are applied to-

gether in the network. Since we are able to deter-

mine which sensor nodes fail to send data in the net-

work, we can efﬁciently switch to a different mode

for every sensor node on demand. The idea is that

the non-failing sensor nodes will continue transmit-

ting data using the No-ACK mode while the failing

sensor nodes are switched to the ACK mode to guar-

antee reliability.

We expect that in the described way, unnecessary

energy consumption and latency caused by the ACKs

in WSNs can be reduced. The adaptability of the pro-

posed protocol makes the network able to combine

the beneﬁts of the energy efﬁciency in No-ACK mode

and the robustness in ACK mode on demand.

4 RESULTS AND DISCUSSIONS

The energy consumption in WSNs basically depends

on the duty cycle and packet size of the sensor nodes,

which is directly associated with their active time. In

fact, the duration of the active time depends on both

the data transfer speed of the network and the type of

protocol used in the network (Hadded et al., 2015).

In this section, we focus only on the latter and com-

pare the results of using ACK, No-ACK and mixed

modes in sensors networks. To discuss the results,

An Adaptive Acknowledgement On-demand Protocol for Wireless Sensor Networks

177

Figure 4: Proposed algorithm of the adaptive acknowledgement on-demand protocol for wireless sensor networks.

we provide network parameters such as packet deliv-

ery ratio (PDR), data throughput, active time and duty

cycle of the sensor nodes. The tests are conducted us-

ing BRIX

modules with widely used existing proto-

cols namely the un-slotted CSMA/CA method imple-

mented in accordance with the IEEE 802.15.4 stan-

dard, TDMA with ACK and TDMA without ACK.

The RF interface of BRIX

is set up to 868 MHz fre-

quency band with 0 dBm transmit power using the

Anaren Integrated Radio 66089 series antenna.

Figure 5: Indoor ofﬁce environment test setup for packet

delivery ratio.

4.1 Packet Delivery Ratio

The test setup created to measure the PDR is done in

an offcie indoor, simulating a typical smart home en-

vironment. The building is mainly constructed using

steel and glass. For the experiment, 6 sensor nodes

were deployed across the building along with a cen-

tral node to retrieve the data from the deployed sensor

nodes (Fig. 5). In this scenario, 3 sensor nodes, (node

1, 2 and 3) are located in the direct line of sight (LOS)

with the central node along the corridor. The other 3

sensor nodes (node 4, 5 and 6) are placed in non-line

of sight (NLOS) locations.

For each test, 1000 packets were transferred with

the maximum payload size of 20 bytes. All tests are

repeated 50 times (Fig. 6). According to the test re-

sults, the average PDR of the nodes in LOS condition

is 100 % in both the ACK and No-ACK mode (Fig.

6). Only node 3 shows a lower PDR (99.5 %) if tested

in No-ACK mode. In the NLOS condition, the aver-

age PDR of node 4 is 100 % in both modes while it is

99.6 % for node 5 when tested without ACK. In gen-

Node1 Node2 Node 3 Node 4 Node 5 Node 6

20%

40%

60%

80%

100%

No-ACK ACK

Sensor Nodes

Figure 6: Comparison of the packet delivery ratio for

TDMA with ACK and without ACK.

SENSORNETS 2017 - 6th International Conference on Sensor Networks

178

TDMA (No-ACK) TDMA (ACK) CSMA/CA

55%

60%

65%

70%

75%

80%

85%

Type of Tested Protocols

Figure 7: Comparison of packet delivery ratio for single

sensor node in congested location 6.

eral, there are no signiﬁcant deviations between the

average PDRs (ACK vs. No-ACK) concerning node

1 to 5. However, differences in the average PDRs be-

tween ACK and No-ACK modes are found for sensor

node 6. In the particular position of node 6, a higher

average PDR is achieved when ACK mode is used

(Fig. 6). In this context, one-time packet retransmis-

sion is allowed in TDMA using ACK mode.

Moreover, the test is conducted with different

schemes to compare the average packet delivery ra-

tio for a single sensor node (Fig. 7), which is placed

in the position of node 6 (Fig. 5). Results show that a

better PDR is reached in TDMA with ACK compared

to TDMA without ACK. The highest PDR is achieved

in CSMA/CA scheme, in which up to three packet re-

transmission are allowed. However, packet losses are

occurred in all of the three tested schemes from this

speciﬁc sensor node’s location.

4.2 Data Throughput

Additionally, we compare the test results of data

throughput for TDMA with ACK, without ACK and

different conditions of mixed ACK and No-ACK

mode (Fig. 8). For all tests, the maximum data

size is 20 bytes. The graph represents the mean data

throughput received at the central node for the whole

Individual ACK, No-ACK and Mixed ACK and No-ACK methods

IDEAL 6_No-ACK

& 0_ACK

5_No-ACK

& 1_ACK

4_No-ACK

& 2_ACK

3_No-ACK

& 3_ACK

2_No-ACK

& 4_ACK

1_No-ACK

& 5_ACK

0_No-ACK

& 6_ACK

Figure 8: Comparison of data throughput in different

schemes at run time period of 30 minutes.

network in 30 minutes. The ideal case (Fig. 8)

refers to the data throughput measured directly be-

tween only two BRIX

modules with maximum pay-

load and no channel access method.

The throughput declines as the number of sensor

nodes that use ACK in the network increases. Thus,

highest data throughput is achieved with TDMA in

No-ACK mode. This value is also closed to the ideal

value (12.83 kbit/s). Lowest throughput is received

when all nodes are using ACK mode (4.8 kbit/s).

The reason that the data throughput of the ACK

mode in TDMA is noticeably lower compared to the

No-ACK mode is that the ACK requires additional

waiting and processing time on both sides of the

transceivers for the ACK signal. The problem of this

ACK time becomes more crucial if the size of the

ACK signal outweighs the data size ifself (or com-

parable to it). As previously stated, this is common

especially in energy constraint low power and low

data rate WSNs in which only small data packages

are transferred.

4.3 Active Time of the Sensor Nodes

The active time of the sensor nodes is directly associ-

ated with the duty cycle. In turn the duty cycle is di-

rectly related to the energy consumption of the whole

network. This is particularly true when the sensor

nodes are transferring data through half duplex anten-

nas. The active time of a sensor node, as explained in

section 3.3, refers to the data transferring time plus the

idle waiting time of ACK signal (in ACK mode) with-

out going to their sleep mode. Therefore, we compare

the minimum and maximum active time of the BRIX

module depending on the protocols that are used in

the test (Fig. 9). In general, the active time for TDMA

in the No-ACK mode is more than bisected compared

to that of the ACK mode. Thus, the life time of the

sensor nodes could be approximately doubled if the

network is running with the No-ACK mode all the

time. The minimum and maximum active times are

TDMA (No-ACK) TDMA (ACK) CSMA/CA

Header Only Max. Payload

2.91

183

Typed of Tested Protocols

Figure 9: Comparison of minimum and maximum active

time for three tested protocols in BRIX

module.

An Adaptive Acknowledgement On-demand Protocol for Wireless Sensor Networks

179

6_No-ACK

& 0_ACK

5_No-ACK

& 1_ACK

4_No-ACK

& 2_ACK

3_No-ACK

& 3_ACK

2_No-ACK

& 4_ACK

1_No-ACK

& 5_ACK

0_No-ACK

& 6_ACK

Header Only Max. Payload

Individual ACK, No-ACK and Mixed ACK and No-ACK Modes

Average Duty Cycle (%)

Figure 10: Comparison of the average duty cycle for ACK,

No-ACK and mixed ACK and No-ACK modes.

measured for the two severe case scenarios:(i) header

part only (no payload) and (ii) maximum data payload

of 20 bytes. The maximum active time of CSMA/CA

is based on the maximum random back-off time of the

handshaking between the request-to-send (RTS) and

clear-to-send (CTS) mechanism of the IEEE 802.15.4

standard.

4.4 Duty Cycle

The average duty cycle of each sensor node (Fig. 10)

is based on the TDMA frame cycle of 200 ms when

6 sensor nodes are deployed in the network. This

means the central node collects the data from each

sensor node in the network every 200 ms. For both the

header only and maximum payload data transmission,

the lowest average duty cycle is found in the No-ACK

mode with 1.4 % and 7 % respectively. Maximum

average duty cycles (header and payload) with 6 %

and 16.5 % are found when the ACK mode is used by

all of the sensor nodes in the network. According to

the test results, it can be concluded that the average

duty cycle of the sensor nodes in the network would

gradually be increased according to the number of the

sensor nodes that are running in ACK mode.

4.5 Summary

According to the test results presented in this section,

a mixture of TDMA with ACK and No-ACK should

be used for energy efﬁcient and reliable data transmis-

sion. For instance, node failure in sensor node 6 (Fig.

5) would cause all the sensor nodes in the network to

switch to ACK mode in normal networking protocol.

However, only node 6 needs to be switched to ACK

using our proposed protocol. Thus, the data through-

put will be 9.14 kbit/s instead of 4.8 kbit/s in which

all the sensor nodes are using ACK mode. Simulta-

neously, the duty cycle of the maximum data payload

will also be reduced from 16.5 % to 8.58 %.

5 CURRENT LIMITATIONS

With our current implementation of the proposed

adaptive acknowledgement on-demand protocol for

WSNs, the maximum number of sensor nodes that can

operate in mixed ACK mode is limited by the total

payload length of the protocol scheme. This is be-

cause the switching of the modes is controlled by the

central node. This is done by individually adding the

sensor node’s identiﬁcation number (IDs) to a broad-

cast packet. However, the scheme is subjected to be

improved in a future implementation.

Moreover, the current switching time between

the energy efﬁcient data transmission with No-ACK

mode to mixtures of ACK and No-ACK mode is from

100 ms up to approximately 1 second. This time de-

pends on the number of sensor nodes in the network

and the active and sleeping time of each individual

sensor node. Again, we tend to improve this issue in

a future implementation.

Furthermore, trafﬁc differentiation between nodes

will be accounted in the future implementation of the

protocol to distinguish trafﬁc classes such as high or

low priority, real-time or best-effort.

6 CONCLUSION AND FUTURE

WORK

In this paper, an adaptive ACK on-demand protocol

for WSNs is presented. The protocol is tested on a

hardware implementation using BRIX

devices. The

impact of ACK signals in WSNs is highlighted by

comparing network parameters such as data through-

put, duty cycle, packet delivery ratio and active time

of the sensor nodes. Two main concepts were de-

ployed: Firstly, the protocol selection was done af-

ter all of the sensor nodes were deployed in a spe-

ciﬁc location in order to minimize the unnecessary us-

age of energy and latency caused by the ACK signal.

Secondly, the data was transmitted in TDMA without

ACK as much as possible in order to achieve maxi-

mum energy and latency efﬁciency. As a result, indi-

vidual node failures can effectively be addressed us-

ing our proposed protocol.

However, the central node is expected to be con-

nected to a continuous or large power source, since

the proposed protocol is mainly focusing on the en-

ergy efﬁciency of the sensor nodes, which are solely

supplied with a battery-based power source.

For future work, a simulation model for the pro-

posed adaptive acknowledgement on-demand proto-

col for WSNs will be designed. The purpose is to fur-

ther explore the impact of ACK in WSNs by varying

SENSORNETS 2017 - 6th International Conference on Sensor Networks

180

the network parameters such as throughput, packet er-

ror rate, duty cycle and active time for both custom

and general designs.

ACKNOWLEDGEMENTS

This work was supported by the Cluster of Ex-

cellence Cognitive Interaction Technology ’CITEC’

(EXC 277) at Bielefeld University, which is funded

by the German Research Foundation (DFG), and

the German Federal Ministry of Education and Re-

search (BMBF) within the project “KogniHome” and

the Leading-Edge Cluster ”Intelligent Technical Sys-

tems OstWestfalenLippe” (it’s OWL), managed by

the Project Management Agency Karlsruhe (PTKA).

Author Cung Lian Sang was supported by German

Academic Exchange Service (DAAD). The authors

are responsible for the contents of this publication.

REFERENCES

Al-Sharbaty, F. S. (2014). Wireless Sensors Network and

Acknowledgement Technique Based ZigBee System.

In 2014 6th International Conference on Computer

Science and Information Technology (CSIT). IEEE.

Chand, T. and Kakria, A. (2015). Comparative Analysis

of a Contention Based (RI-MAC) and TDMA Based

(ATMA) MAC Protocols for Wireless Sensor Net-

works. In SENSORS, 2015 IEEE. IEEE.

De Oliveira, R. and Braun, T. (2005). A Dynamic Adap-

tive Acknowledgment Strategy for TCP over Multi-

hop Wireless Networks. In Proceedings IEEE 24th

Annual Joint Conference of the IEEE Computer and

Communications Societies. IEEE.

Gama,

O., Carvalho, P., and Mendes, P. (2010). A Time-

slot Scheduling Algorithm for E-health Wireless Sen-

sor Networks. In 12th IEEE International Confer-

ence on e-Health Networking Applications and Ser-

vices (Healthcom). IEEE.

Hadded, M., Muhlethaler, P., Laouiti, A., Zagrouba, R., and

Saidane, L. A. (2015). TDMA-Based MAC Protocols

for Vehicular Ad Hoc Networks: A Survey, Qualita-

tive Analysis, and Open Research Issues. In IEEE

Communications Surveys & Tutorials. IEEE.

IEEE (2011). Part 15.4: Low-Rate Wireless Personal Area

Networks (LR-WPANs). IEEE Standards Association,

New York, revision of ieee std 802.15.4-2006 edition.

Liu, Z., Liu, B., Chen, C., and Chen, C. W. (2016). An

Energy-efﬁcient and QoS-effective Resource Alloca-

tion Scheme in WBANs. In 2016 IEEE 13th Interna-

tional Conference on Wearable and Implantable Body

Sensor Networks (BSN). IEEE.

Montiel, N. L. and C

ardenas, L. M. R. (2014). Data Routing

Protocol Using TDMA for a Wireless Sensor Network

with Structured Topology for a Noisy Environment.

In 2014 International Conference on Mechatronics,

Electronics and Automotive Engineering (ICMEAE).

IEEE.

Scheuermann, B., Lochert, C., and Mauve, M. (2008). Im-

plicit Hop-by-Hop Congestion Control in Wireless

Multihop Networks. In Ad Hoc Networks. Elsevier.

Shu, F., Sakurai, T., Zukerman, M., and Vu, H. L. (2007).

Packet Loss Analysis of the IEEE 802.15. 4 MAC

without Acknowledgements. In IEEE communica-

tions letters. IEEE.

Takamori, M. and Yamao, Y. (2015). Enhancing Through-

put of Star Topology Sensor Network by Group Ac-

knowledgement Method and MCR SS-CSMA/CA. In

2015 21st Asia-Paciﬁc Conference on Communica-

tions (APCC). IEEE.

Zehe, S., Grosshauser, T., and Hermann, T. (2012). BRIX

- An Easy-to-use Modular Sensor and Actuator Pro-

totyping Toolkit. In 2012 IEEE International Con-

ference on Pervasive Computing and Communications

Workshops (PERCOM Workshops). IEEE.

An Adaptive Acknowledgement On-demand Protocol for Wireless Sensor Networks

181