Evaluation of Multi-Channel Communication for an Outdoor Industrial

Wireless Sensor Network

Ruan D. Gomes

, Emerson B. Gomes

, Iguatemi E. Fonseca

, Marcelo S. Alencar

and Cesar Benavente-Peces

Federal Institute of Para

ıba, Guarabira, 58051-900, Brazil

Federal University of Para

ıba, Jo

ao Pessoa, 58058-600, Brazil

Federal University of Campina Grande, Campina Grande, CEP: 58401-490, Brazil

Universidad Politecnica de Madrid, Madrid, Spain

Keywords:

Industrial Wireless Sensor Networks, Multi-channel Communication, Wireless Channel Characterization,

Outdoor Industrial Environment.

Abstract:

This paper describes an experimental study which investigates relevant properties of multi-channel wireless

communications in an outdoor industrial environment. A testbed of IEEE 802.15.4 radios was developed in

order to evaluate the performance of the 16 channels deﬁned by the standard, at all the nodes, simultaneously.

From the collected data, some relevant facts are discussed, such as the spatial variations in channel quality,

the differences in the characteristics of different channels, the link asymmetry, and the non-stationary charac-

teristics of the channel. The possible problems that can arise in the deployment of industrial wireless sensor

networks, based on the characteristics of the standards developed for this type of network, are described, as

well as some possible solutions.

1 INTRODUCTION

The use of Industrial Wireless Sensor Networks

(IWSN) to implement monitoring and control systems

has some advantages, such as low cost and high ﬂex-

ibility to reconﬁgure the network. However, it is nec-

essary to deal with typical problems of wireless net-

works, such as noise, electromagnetic interference,

fading and high attenuation, due to the presence of

many objects and obstructions. Many industrial en-

vironments also present characteristics that make the

wireless channel non-stationary for long time peri-

ods (Agrawal et al., 2014).

Another problem is the link asymmetry. Some

protocols use acknowledgement per packet and, in

this case, it is necessary to guarantee a good quality of

communication in the two directions of the link. Spa-

tial variations in the channel quality can also occur in

IWSN. In (Watteyne et al., 2010), a coherence length

of 5.5 cm was found for IEEE 802.15.4 radios operat-

ing in the 2.4 GHz band. Hence, two nodes positioned

at a distance more than 5.5 cm apart from each other,

and using the same channel, can be considered uncor-

related, and thus the channel can present a high qual-

ity for one node, and a low quality for the other.

Some standards have been proposed in the last

years with a focus on industrial applications, such

as the WirelessHART and the ISA100.11a, which are

based on the physical layer of the IEEE 802.15.4 stan-

dard, but deﬁne their own MAC layer based on Time

Division Multiple Access (TDMA), to avoid colli-

sions, and reduce the power consumption. They also

use frequency hopping and blacklisting, to mitigate

the problems related to interference and fading.

More recently, the IEEE 802.15.4e standard was

released, which proposes solutions for applications

that require high reliability (e.g. industrial applica-

tions) (Guglielmo et al., 2016). Five modes of op-

eration are deﬁned, but only the Time-Slotted Chan-

nel Hopping (TSCH), Deterministic and Synchronous

Multi-Channel Extension (DSME), and Low Latency

Deterministic Network (LLDN) modes have been ex-

plored in the literature, until now. In general, the

modes are based on TDMA or frequency hopping to

reduce collisions and mitigate the effects of interfer-

ence and fading.

Gomes, R., Gomes, E., Fonseca, I., Alencar, M. and Benavente-Peces, C.

Evaluation of Multi-Channel Communication for an Outdoor Industrial Wireless Sensor Network.

DOI: 10.5220/0006609200600066

In Proceedings of the 7th International Conference on Sensor Networks (SENSORNETS 2018), pages 60-66

ISBN: 978-989-758-284-4

Even these new protocols deﬁne mechanisms to

deal with the unreliability problems of IWSN, it is

necessary to analyze the characteristics of the multi-

channel communication in such environments, in or-

der to properly deploy the network. For exam-

ple, when using channel hopping, the nodes usually

switch to a new channel before each transmission.

However, if a proper management of the blacklist is

not made, the network performance can be signiﬁ-

cantly degraded (Grsu et al., 2016). Problems due to

the spatial variations in channel quality can also affect

the performance of beacon-based protocols.

In this paper, the characteristics of the 16 chan-

nels, deﬁned by the IEEE 802.15.4 standard, are an-

alyzed, in an outdoor industrial environment, and for

eight different links. Based on the experimental re-

sults, possible problems that can arise in the deploy-

ment of IWSN, and some possible solutions are de-

scribed. The parameters of the log-normal shadow-

ing model for the environment were also determined.

Some studies have been performed in outdoor indus-

trial environments (Boano et al., 2010) to analyze the

impact of environmental aspects (e.g. temperature)

on link quality. The novel contribution of this paper

is the detailed analysis of the multi-channel commu-

nication in an outdoor industrial environment, which

may be important to design new techniques and pro-

tocols, as well as, more accurate simulation and theo-

retical models.

1.1 The Wireless Channel in Industrial

Environments

The industrial environment usually contains metallic

and mobile objects, such as robots, cars and people.

This inﬂuences both the large-scale and small-scale

fading. The power of the received signals depends

on the transmission power, the antennas gains, the

distance between transmitter and receiver and the ef-

fects caused by the environment. Even with the same

values for the aforementioned parameters, there is a

variation in the mean received power, depending on

the place where the measurement is performed, which

is known as log-normal shadowing. The log-normal

shadowing model has been used to model the large-

scale path loss and shadowing in industrial environ-

ments (Tanghe et al., 2008).

Besides path loss and shadowing, it is also nec-

essary to analyze the small-scale channel fading due

to rapid changes in the multipath proﬁle of the en-

vironment, which is caused by the movement of ob-

jects around the receiver and transmitter. Experi-

ments demonstrated that, in industrial environments,

the temporal attenuation follows a Rice distribution.

In industrial environments the K factor of the Rice

distribution has a high value. For the experiments de-

scribed in (Tanghe et al., 2008), in industrial environ-

ments, K presented values between 4 dB and 19 dB,

while in ofﬁce environments, values between -12 dB

and -6 dB were reported, as discussed in (Tanghe

et al., 2008). This can be explained by the open nature

of industrial buildings and the large amount of reﬂec-

tive materials. Thus, there are many time-invariant

rays and only a small part of the multipath proﬁle is

affected by moving objects.

The IEEE 802.15.4 standard deﬁnes sixteen chan-

nels in the 2.4 GHz band, with 2 MHz of bandwidth,

and channel spacing of 5 MHz. Thus, the chan-

nels are highly uncorrelated. Experiments described

in (Amzucu et al., 2014) have found that changing

the communication channel can lead up to 30 dB dif-

ference in the received power, in an ofﬁce environ-

ment. Varga et al. (Varga et al., 2016) performed ex-

periments for a short range, in an environment with-

out multipath, and with line-of-sight. In that experi-

ment, differences up to 10 dB were observed for some

channels. Thus, besides the variation in shadowing

observed depending on the place that the nodes are

positioned, there is also a variation in shadowing re-

garding the different channels. In the experiments de-

scribed in (Gomes et al., 2017), differences of up to

15 dB were found for different channels in an indoor

industrial environment, but only one link was ana-

lyzed.

In this paper, the aspects that inﬂuences the chan-

nel characteristics are discussed, based on experi-

ments performed in an outdoor industrial environ-

ments, and considering eight links simultaneously, to

analyze the temporal, spatial, and frequency varia-

tions in channel quality.

2 EXPERIMENT

METHODOLOGY

The sensor nodes used in the experiment include

an MRF24J40MA transceiver, with a transmission

power of 0 dBm, a PCB antenna with a gain

of 2.09 dBi, and a PIC18F46J50 micro-controller.

Eight sensor nodes (1 to 8), and a coordinator (9),

were placed in an outdoor industrial environment

(Fig. 1(a)), according to the schematic shown in

Fig. 1(b). The industrial unit is a water treatment and

injection station, which treats the water that comes

together with the oil from onshore oil ﬁelds and send

it, pressurized, to a group of platforms placed about

25 km from the station. During the experiments, the

station was operating normally, and the sensor nodes

Evaluation of Multi-Channel Communication for an Outdoor Industrial Wireless Sensor Network

Figure 1: (a) Environment where the experiments were per-

formed. (b) Schematic.

were placed alongside wired sensors that are currently

installed in the unit.

To allow the nodes to communicate using all chan-

nels, and without collisions, a protocol based on

TDMA and channel hopping was implemented. In the

protocol, the medium access occurs based on a slot-

frame structure, which repeats continuously, similar

to the slotframe deﬁned on the TSCH protocol, but

with the use of beacons, transmitted by the coordina-

tor to synchronize the end-nodes in each cycle. The

temporal structure of the slotframe is shown in Fig-

ure 2.

Figure 2: Slotframe structure of the implemented protocol.

The slotframe repeats continuously and is com-

posed by 10 slots. In the ﬁrst slot the coordinator

transmits the beacon in broadcast, and waits to receive

data packets that are transmitted by the end-nodes in

the following eight time slots (S1 to S8). There is an

inactive interval in the end of the slotframe, that is

used by the nodes to switch the channel and wait the

next slotframe.

When an end-node receives a beacon from the co-

ordinator, it waits until the time interval allocated to

it and performs the transmission of a data packet to

the coordinator. Each slot has a duration of 100 ms.

This protocol was developed to allow the characteri-

zation of the multi-channel communication for mul-

tiple links simultaneously, but it was not developed

taking under consideration any particular application.

In each slotframe a different channel is used, se-

quentially. To accommodate the use of channel hop-

ping in the transmission of the beacons, it is necessary

to have a mechanism to maintain the network syn-

chronized in case of failures during the reception of

a beacon. To do this, a timer is used in the end-nodes

to identify that a beacon has been lost. The coordina-

tor sends a new beacon for each 1 s, thus the timer is

conﬁgured to expire after 1.1 s. If a node receives a

new beacon before the timer expires, the timer is re-

seted. Otherwise, the node switches the channel, and

waits for the next beacon, which maintains the syn-

chronization.

After the reception of a beacon, the end-nodes ob-

tain the Received Signal Strength Indication (RSSI)

of the beacon, and transmit it back to the coordinator.

For each received packet at the coordinator, the RSSI

of the packet, as well as the RSSI of the beacon, sent

by the end-node, are uploaded to a computer through

a serial port. Thus, it is possible to analyze the spa-

tial variations in the channel quality, and asymmetry,

for all links. Even the individual RSSI samples are

obtained in different moments for the different nodes

and channels, due to the TDMA protocol, with the ac-

quisition of many samples over time, it is possible to

obtain the mean received power and the standard devi-

ation, and compare the characteristics of the different

channels for the different links.

Two experiments were performed, in two days,

with the nodes positioned in the same place. The net-

work operated for about 3 h and 10 h, in the ﬁrst and

second days, respectively. The values of RSSI pro-

vided by the MRF24J40 transceiver varies between 0

and 255. For packets received with power between

-94 dBm (transceiver sensitivity) and -90 dBm, the

RSSI is equal to zero. However, despite this limi-

tation, it was possible to analyze the differences in

the characteristics of all channels, the spatial varia-

tions, and the non-stationary behavior of the wireless

channel, and drawn remarkable conclusions about the

SENSORNETS 2018 - 7th International Conference on Sensor Networks

Figure 3: Spatial variations in the channel quality for different nodes, in the ﬁrst day of experiment.

multi-channel communication in outdoor industrial

environments.

3 RESULTS

Fig. 3 shows the mean received power, and the stan-

dard deviation, for each end-node, considering the ex-

periment performed in the ﬁrst day. The mean re-

ceived power varies signiﬁcantly, even for the adja-

cent channels and for the same end-node. For ex-

ample, for Node 1, the differences for some chan-

nels were higher than 10 dB (e.g. Ch 20 and Ch 25).

For Nodes 4 and 5, which were positioned in a place

without Line-Of-Sight (LOS) (see Fig. 1(b)), no com-

munication can be set, for example, when using Ch

22, due to a deep fading problem in the channel. All

channels shown a low quality for Node 5, but for

Node 4 some channels presented high quality, such

as the Ch 17. Deep fading problems have also oc-

curred for some other nodes and channels, in which

the number of packets received was very low.

From Fig. 3 it is also possible to analyze the spa-

tial variations in the quality of the channels. The val-

ues of received power for Nodes 3 and 7 are analyzed

in detail for two different channels. These nodes were

positioned at nearly the same distance to the coordina-

tor, and with a 1.6 m of difference in the height. The

Ch 17 presented a high quality for Node 3, but the

quality was signiﬁcantly lower for the Node 7. On

the other hand, the Ch 21 presented a high quality

for Node 7, but a low quality for Node 3. The re-

ception power in the two directions of the links are

shown. There is a high correlation between the re-

ceived power in both directions of the links, but with

a small difference in the mean values. When the re-

ceived power is near to the sensitivity threshold of the

transceiver, such as is the case of Ch 21 for Node 3,

this small difference can provoke an asymmetry in the

link quality.

Hence, it is difﬁcult to guarantee a good Quality of

Service (QoS) for all nodes when only one channel is

used in the whole network, such as in the MAC proto-

cols deﬁned by the IEEE 802.15.4 standard. Even for

the new standards deﬁned for IWSN, some problems

can arise due to the spatial variations in channel qual-

ity. In the LLDN mode, TDMA is used to avoid colli-

sions, with a star topology, to achieve very low laten-

cies (Anwar et al., 2016). However, only one channel

is used for all end-nodes. One possible solution is the

use of multiple sink-nodes, using different channels.

When the channel being used by an end-node starts

to present low quality, that node can switch to another

channel and communicate with a different sink. How-

ever, some mechanism to estimate the link quality in

real-time (Gomes et al., 2017), and a speciﬁc syn-

chronization mechanism needs to be developed. The

protocol described in (Patti and Bello, 2016) uses a

tree topology and multi-channel communication for

LLDN networks, with adaptive channel selection, but

the same channel is allocated to all nodes in the sub-

network, and spatial variations in channel quality can

also occur inside the same sub-network.

Even for protocols that use channel hopping or

channel adaptation, some problems may arise. For

example, the TSCH, WirelessHART, and ISA100.11a

Evaluation of Multi-Channel Communication for an Outdoor Industrial Wireless Sensor Network

Figure 4: Comparative results between the two days of experiment.

standards use TDMA and channel hopping. In this ap-

proach, all the channels can be used by the end-nodes

to perform communication. However, the blacklist

needs to be properly managed in order to achieve a

good QoS in the network. In (Du and Roussos, 2013)

it was observed that the larger the size of the black-

list, the better the communication performance. This

result corroborates with the results presented in (Grsu

et al., 2016). However, this type of behavior only oc-

curs if an adequate monitoring of the quality of the

channels is performed, in order to properly conﬁgure

the blacklist.

One problem is that, when a channel is black-

listed, all the nodes stop using that channel. In the

result shown in Fig. 3, Ch 21 presented a low quality

for four end-nodes and could be put on the blacklist.

However, this channel is the one that presents the best

quality for Node 7, and also presents good quality for

Nodes 1 and 6. Thus, the QoS for these nodes can de-

crease once this channel is put on the blacklist. When

the quality of the channel is affected by external in-

terference, as considered in (Du and Roussos, 2013),

putting a channel in the blacklist for all network can

be a good solution, but the challenge is higher when

spatial variations in channel quality, due to multipath

problems, affect the links.

The DSME mode employs channel hopping or

channel adaptation, during the contention free peri-

ods. When using the channel adaptation, a pair of

nodes can communicate using the same channel for

a long time period, and a channel switch only occurs

when the channel in use starts to present low quality.

Thus, it is possible to deal with the spatial variations

in the quality of the channels, since the decision about

the channel to be used can be made based on the qual-

ity of a speciﬁc link between a given pair of nodes.

The implementation of this procedure is not deﬁned

by the standard (Guglielmo et al., 2016). The DSME

networks use beacon packets, transmitted in broad-

cast using a single channel. Sometimes it is difﬁcult

to pick one channel that presents good quality for all

nodes in the network. Deep fading problems can also

occur, and some nodes can remain disconnected for a

long time. While the use of channel adaptation can be

a good solution for unicast data packets, channel hop-

ping can be a good solution for packets transmitted in

broadcast.

Fig. 4 shows a comparison between the results ob-

tained in the two days. Fig. 4(a) shows the mean re-

ceived power, and the variance for all channels. The

variance was high in all channels, due to the differ-

ences in the channel characteristics for the different

links. Fig. 4(b) shows the mean received power, and

the variance, for the eight different links, consider-

ing the 16 channels. There is also a signiﬁcant vari-

ance, due to the differences in the characteristics of

the different channels in each link. Fig 4(c) shows the

results for a speciﬁc node (Node 2) and for all chan-

nels in both days. It is worthy to notice that while

some channels had an increase in quality, the quality

of other channels decreased signiﬁcantly in the sec-

ond day. For example, the Ch 18 presented a good

quality in the ﬁrst day for Node 2, but presented a

deep fading problem during the second day. Fig 4(d)

shows the results for a speciﬁc channel (Ch 21), and

for the eighth different links. It is possible to notice

SENSORNETS 2018 - 7th International Conference on Sensor Networks

that the characteristics of the channels vary differently

for the different nodes. For example, the Ch 21 pre-

sented a high quality for the Node 6 in the ﬁrst day,

but a low quality in the second day. On the other hand,

it presented a higher quality on the second day for the

Node 2.

Some abrupt variations in the reception power for

some nodes and channels were also observed during

the second day (Fig. 5). This behavior was also ob-

served in (Agrawal et al., 2014) for an indoor indus-

trial environment. The quality of the Ch 24 decreased

after some time for Node 7, while at the same time the

channel showed an increase in its quality for Node 6.

Again, a high correlation between the two directions

of the links was observed, but with a small difference

in the mean value of each direction of the link.

Figure 5: Abrupt change in channel characteristics.

The chart in Fig. 6 shows the path loss (L(d)) for

a distance (d) between transmitter and receiver. The

values obtained for Node 1 were used as reference

= 16.5 m). From this experiment, the path loss

exponent (n), the shadowing deviation (σ), and L(d

)

were obtained, to be applied in the log-normal shad-

owing model. Fig. 6 shows the curves of the model

for three different scenarios: with all nodes, with the

LOS nodes (Nodes 3, 6, 7 and 8), and with the NLOS

nodes (Nodes 2, 4, and 5). Node 1 was considered in

both cases as the reference. Table 1 shows the param-

eters for the three scenarios. These values can be used

to simulate outdoor IWSN.

To allow an accurate simulation, it is important

to consider all the aspects and conclusion discussed

in this paper, and also nodes with LOS and NLOS,

with different parameters for the path loss and shad-

owing. Also, the level of shadowing in each channel

for the different links need to be modiﬁed over time.

Sometimes abrupt changes can occur in the charac-

teristics of the channel, and the modiﬁcations occur

differently for the different channels and nodes, and

the protocols for IWSN need to be capable to dealing

with these modiﬁcations to maintain a good QoS over

time.

Table 1: Parameters for the log-normal shadowing model.

Path Loss

Exponent (n)

Shadowing

Deviation (σ)

L(d

)

All Nodes 2.00 4.53 dB 81.182 dB

LOS Nodes 2.43 4.54 dB 78.351 dB

NLOS Nodes 4.03 4.98 dB 80.352 dB

4 CONCLUSIONS

This paper describes a set of experiments to evalu-

ate and characterize the performance of multi-channel

communications in an outdoor IWSN. Relevant char-

acteristics of the wireless channel were described,

based on the experimental results. Some problems

that can occur in the deployment of an IWSN, as well

as some possible solutions, are discussed in the paper,

considering the characteristics of the wireless channel

in the environment under study, and the characteris-

tics of the protocols that are used to implement the

IWSN. The parameters of the log-normal shadowing

model were also obtained. The information contained

in this paper can be used to allow the implementation

of new techniques and protocols for IWSN, as well as

more accurate simulation models.

Based on the characteristics of the wireless chan-

nel that were observed in the experiments, some fu-

ture works can be outlined, such as the design and

implementation of mechanisms for dynamic conﬁg-

uration of the blacklist in protocols that use fre-

quency hopping, to deal with the time and spatial

variations in the quality of the channels. The use

of link quality estimators to monitor the quality of

the links continuously can be useful to improve the

blacklist management. Other alternative is the use of

channel adaptation mechanisms, in which the nodes

use only one channel to communicate, but the chan-

nel is changed whenever the quality of the channel

in use is below a certain threshold. All these as-

pects will be investigated in future works, as well

as new experiments in other types of outdoor in-

dustrial environments will be performed. The data

generated during the experiments are available at:

https://github.com/ruandg/ExpIWSNoutdoor.

Evaluation of Multi-Channel Communication for an Outdoor Industrial Wireless Sensor Network

Figure 6: Relation between the path loss and the distance between transmitter and receiver.

ACKNOWLEDGEMENTS

The authors would like to thank the support of the

Institute for Advanced Studies in Communications

(Iecom), the Brazilian Council for Research and De-

velopment (CNPq), the Coordination for the Improve-

ment of Higher Education Personnel (CAPES), and

Petrobras.

REFERENCES

Agrawal, P., Ahl

en, A., Olofsson, T., and Gidlund, M.

(2014). Long term channel characterization for en-

ergy efﬁcient transmission in industrial environments.

IEEE Trans. on Communications, 62(8):3004–3014.

Amzucu, D. M., Li, H., and Fledderus, E. (2014). Indoor

radio propagation and interference in 2.4 ghz wireless

sensor networks: Measurements and analysis. Wire-

less Personal Communications, 76:245–269.

Anwar, M., Xia, Y., and Zhan, Y. (2016). Tdma-based

ieee 802.15.4 for low-latency deterministic control ap-

plications. IEEE Trans. on Industrial Informatics,

12(1):338–347.

Boano, C. A., Tsiftes, N., Voigt, T., Brown, J., and Roedig,

U. (2010). The impact of temperature on outdoor in-

dustrial sensornet applications. IEEE Trans. on Indus-

trial Informatics, 6(3):451–459.

Du, P. and Roussos, G. (2013). Spectrum-aware wireless

sensor networks. In 2013 IEEE 24th Annual Inter-

national Symposium on Personal, Indoor, and Mobile

Radio Communications (PIMRC), pages 2321–2325.

Gomes, R. D., Queiroz, D. V., Filho, A. C. L., Fonseca,

I. E., and Alencar, M. S. (2017). Real-time link qual-

ity estimation for industrial wireless sensor networks

using dedicated nodes. Ad Hoc Networks, 59:116 –

133.

Grsu, M., Vilgelm, M., Zoppi, S., and Kellerer, W. (2016).

Reliable co-existence of 802.15.4e tsch-based wsn

and wi-ﬁ in an aircraft cabin. In 2016 IEEE Inter-

national Conference on Communications Workshops

(ICC), pages 663–668.

Guglielmo, D. D., Brienza, S., and Anastasi, G. (2016).

{IEEE} 802.15.4e: A survey. Computer Communi-

cations, 88:1 – 24.

Patti, G. and Bello, L. L. (2016). A priority-aware multi-

channel adaptive framework for the ieee 802.15.4e-

lldn. IEEE Transactions on Industrial Electronics,

63(10):6360–6370.

Tanghe, E., Joseph, W., Verloock, L., Martens, L., Capoen,

H., Herwegen, K. V., and Vantomme, W. (2008). The

industrial indoor channel: Large-scale and temporal

fading at 900, 2400, and 5200 mhz. IEEE Trans. on

Wireless Communications, 7:2740–2751.

Varga, L.-O., Heusse, M., Guizzetti, R., and Duda, A.

(2016). Why is Frequency Channel Diversity so Ben-

eﬁcial in Wireless Sensor Networks? In IFIP Wireless

Days, Toulouse, France. IFIP.

Watteyne, T., Lanzisera, S., Mehta, A., and Pister, K. S. J.

(2010). Mitigating multipath fading through channel

hopping in wireless sensor networks. In 2010 IEEE

International Conference on Communications, pages

1–5.

SENSORNETS 2018 - 7th International Conference on Sensor Networks