Future Research Challenges in Wireless Sensor and

Actuator Networks Targeting Industrial Automation

Johan Åkerberg Mikael Gidlund Mats Björkman
ABB AB ABB AB Mälardalen University
Corporate Research Corporate Research School of Innovation, Design, and Technology
Email: [email protected] Email: [email protected] Email: [email protected]

Abstract—A growing trend in the automation industry is detail many recent and emerging topics in general for wireless
to use wireless technologies to reduce cable cost, deployment industrial communications. However, many of the research
time, unlocking of stranded information in previously deployed opportunities foreseen are too general in order to address
devices, and enabling wireless control applications. Despite a
huge research effort in the area of wireless sensor networks the urgent needs for process automation domain. Akyidiz
(WSNs), there are several issues that have not been addressed and Wang present a detailed investigation of current state-
properly such that WSNs can be adopted properly in the process of-art protocols and algorithms for wireless mesh networks
automation domain. [5]. Furthermore, they outline some open research issues in
This article presents the major requirements for typical all protocol layers. However, their survey is targeting more
applications in process automation and we also aim to outline
the research direction for industrial wireless sensor networks consumer applications, where the requirements are different
(IWSNs) in industrial automation. The major issues that need to compared to large scale industrial manufacturing. Yick et al.
be addressed are safety, security and availability before industrial [6] present a similar but updated overview of the work by
wireless sensor networks will be adopted in full scale in process Akyidiz and Wang. Zheng describes very briefly the current
automation. situation of IWSN and ongoing standardization activities [7].
Hou and Bergmann present a brief survey of design require-
I. I NTRODUCTION
ments for condition monitoring using commercial industrial
Reliable wireless communications systems and particularly wireless sensor networks (IWSNs) systems [8]. Nevertheless,
industrial wireless sensor and actuator networks (WSANs) are their article does not discuss several important requirements
removing the physical and economical barriers that previously such as safety, security, and availability. Gungor and Hancke
made it difficult or impossible to access many types of present a brief overview of technical challenges and design
information in the process industry. The main concerns about principles in terms of hardware and software development,
reliability, security, and integration along with the lack of system architectures, and protocols for IWSNs [9]. Specif-
device interoperability have hampered the deployment rate. ically, radio technologies, energy-harvesting techniques, and
There exists a lot of work on employing IEEE 802.15.4 cross-layer designs for IWSNs are discussed. As can be seen,
and ZigBee in process automation. However, ZigBee seems there exist some good research work outlining challenges for
unsuitable for this application field as it has not been specifi- wireless sensor networks in industrial automation but most of
cally designed for reliable real-time cyclic communication [1]. them lack discussions about issues related to safety, security
One of the major problems with using the complete IEEE and how to handle actuators which are the main concerns
802.15.4 standard is for large networks where synchroniza- in process automation today. Routing is an important area of
tion is a huge problem. WirelessHART [2], the first open research with respect to IWSN. Heo et al. [10] address routing
and interoperable wireless communication standard designed with respect to real-time performance and energy awareness.
specifically for real-world applications in process automation, This article presents typical requirements from process
was approved and released in 2007. ISA 100.11a [3] is likely automation for some different applications intended to be used
to become a standard for process and factory automation. Both in industrial wireless sensor and actuator networks. Given
WirelessHART and ISA 100.11a adopt the complete IEEE these requirements, it obvious that today’s commercial WSN
802.15.4 PHY layer, but they propose a new MAC layer which are not tailored for the needs of industrial automation, since
combines TDMA and channel hopping to control access to the they are mostly influenced upon requirements derived from
network. Both WirelessHART and ISA 100.11a are mainly the consumer market or other applications. Furthermore, with
targeting applications such as condition monitoring which has the requirements for process automation in mind we describe
quite relaxed requirements on latency. However, for more time the major challenges that need to be solved in order to utilize
critical applications it is likely that some improvements in the WSANs to the extent the market foresee.
current standards are needed. The rest of this paper is organized as follows. In Section
There exists a significant research work done with respect II, we discuss the main drivers and benefits of using wireless
to wireless mesh networks in general. Willig [4] discuss in sensor networks in industrial environments and highlight some

978-1-4577-0434-5/11/$26.00 ©2011 IEEE 410

 Fig. 1. Future wireless infrastructure and its applications in the process automation domain.

emerging wireless applications. Section III briefly describes of IWSN, we are able to unlock stranded information in
the process automation domain and we define the most im- instruments, gather information from where it previously
portant requirements for industrial WSANs with respect to has been economically unfeasible, such that the process
process automation. In Section IV, we outline and discuss can be enhanced with respect to quality and quantity, and
in detail the major research challenges for industrial WSANs reducing the possibility of mechanical failures. Further-
becoming the huge success the industry anticipate. Finally, the more, another advantage is that IWSN enables temporary
paper is concluded in Section V. measurements of certain process values and quality indi-
cators without installation of additional wires.
II. T HE B ENEFITS OF I NDUSTRIAL W IRELESS S ENSOR
∙ Emerging Applications - With the advent of wireless
AND ACTUATOR N ETWORKS
infrastructure and IWSN in process automation several
Wireless technology and especially wireless sensor net- new wireless applications are emerging such as empow-
works has the potential to contribute significantly in areas such ering mobile workers, location of assets, safety muster-
as cable replacement, mobility, flexibility and scalability. It ing, integration on nontraditional signals such as video,
offers competitive advantages such as lower life-cycle cost bridging remote or isolated control systems, enabling
and reducing connector failures which is one of the most wireless control applications [11], and allow connectivity
common reliability problems. Below we summarize the major for equipment that is “sealed-for-life”.
advantages of industrial wireless sensor networks. ∙ Availability - Industrial applications require availability
∙ Cost - With capital at a premium, process manufactures and determinism in the automation systems to avoid seri-
are looking for quick investments that cost little and save ous consequences such as injury, explosions, and material
even more. The major incitement for adoption of IWSN losses. Industrial wireless sensor networks can offer built
in process automation (all automation business) is that in redundancy and capabilities for anticipatory system
they are easier and less costly to install than traditional maintenance and failure recovery. This could for example
wired systems. Consider a green field installation today, be achieved by designing meshed networks where there
it cost roughly $200 per meter to install wires in an always exist available links to the control system.
ordinary process plant, and approximately $1000 per
meter in offshore installations. As wires age they crack III. IWSN R EQUIREMENTS
or fail. Furthermore, inspecting, testing, troubleshooting, In this section we highlight some important requirements
and upgrading wires require time, labor, and materials. that need to be met for a successful large scale deployment
∙ Flexibility - Many secondary process variables have long of IWSNs in process automation and discrete manufacturing.
gone unmeasured, and expensive pieces of critical rotat- Some typical examples of process automation industries are:
ing equipment remain non-instrumented. With the advent oil and gas, mining, steel, pulp and paper to mention some.

411
 TABLE I
T YPICAL R EQUIREMENTS FOR I NDUSTRIAL W IRELESS S ENSOR AND ACTUATOR N ETWORKS IN THE P ROCESS AUTOMATION D OMAIN

Sensor Network Applications Delay Range Battery Lifetime Update Frequency Security level
Monitoring and supervision
Vibration sensor 𝑠 100 𝑚 3 years sec - days low
Pressure sensor 𝑚𝑠 100 𝑚 3 years 1 sec low
Temperature sensor 𝑠 100 𝑚 3 years 5 sec low
Gas detection sensor 𝑚𝑠 100 𝑚 3 years 1 sec low
Closed loop control
Control valve 𝑚𝑠 100 𝑚 >5 years 10 − 500 𝑚𝑠 medium
Pressure sensor 𝑚𝑠 100 𝑚 >5 years 10 − 500 𝑚𝑠 medium
Temperature sensor 𝑚𝑠 100 𝑚 >5 years 500 𝑚𝑠 medium
Flow sensor 𝑚𝑠 100 𝑚 >5 years 10 − 500 𝑚𝑠 medium
Torque sensor 𝑚𝑠 100 𝑚 >5 years 10 − 500 𝑚𝑠 medium
Variable speed drive 𝑚𝑠 100 𝑚 >5 years 10 − 500 𝑚𝑠 medium
Interlocking and Control
Proximity sensor 𝑚𝑠 100 𝑚 >5 years 10 − 250 𝑚𝑠 medium
Motor 𝑚𝑠 100 𝑚 >5 years 10 − 250 𝑚𝑠 medium
Valve 𝑚𝑠 100 𝑚 >5 years 10 − 250 𝑚𝑠 medium
Protection relays 𝑚𝑠 100 𝑚 >5 years 10 − 250 𝑚𝑠 medium

The main characteristic that groups them together is that the ∙ Interlocking and control - normally the major part of
products are produced in a continuous manner, i.e. the oil control applications in process control require discrete
is produced in a continuous flow. In discrete manufacturing, signaling. For example, before a start command can be
the products are produced in discrete steps, i.e. the products issued to a machine, several start conditions have to be
are assembled together using sub assemblies or single compo- fulfilled. A machine might have start, stop, and safety-
nents. Typical examples of discrete manufacturing industries interlocks. Therefore, interlocking and control signaling
are automotive, medical, and the food industries. Discrete are sensitive to delays.
manufacturing heavily relies on robotics and belt conveyors The reason for the separation of closed loop control and
for assembly, picking, welding, and palletizing. To generalize, interlocking and control is that when closed control is used,
discrete manufacturing normally have stricter requirements the controller can compensate for delays and retransmissions
with respect to latency and real-time requirements compared to without major degradation of control performance. On the
process automation. However, as always there are cases when other hand a process interlock, that should for instance stop a
this general assumption is not true. The main reason for this machine, is less tolerant for delays and retransmissions as the
generalized assumption is that in order to pick, assemble, or consequences of delays are most likely to introduce spurious
palletize at high speed, the latency, refresh rates, and real-time problems in the production or even dangerous situations.
requirements are stricter compared to a tank level control in
process automation to achieve the required production quality. IV. R ESEARCH C HALLENGES
However, in this paper we focus of process automation, while
As previously introduced there are advantages of using
keeping in mind that requirements might be even stricter for
IWSNs in process automation. However, there are important
discrete manufacturing.
requirements that cannot be met in an efficient way today, or
In Table I, the different types of applications are grouped cannot be met at all. In this section we discuss some major
and listed in three sub-categories: issues that need to be addressed in order to achieve a large
∙ Monitoring, supervision - this sub-category collects dif- scale deployment of IWSNs. In order to be competitive and
ferent types of sensors that provide diagnostics and super- cost efficient, we have to consider that the IWSNs needs to
vision that normally can be pre-processed and transmitted solve all requirements that is met today using wired field-
and updated on a period time from 1 second and more. buses. The end users will not afford to invest and maintain
This information is generally not sensitive for packet parallel infrastructures. Secondly, if we cannot meet most of
losses and jitter as it is used for supervision and condition the requirements with IWSNs we need to wire up the field
monitoring. However, in some cases data consistency equipment, which is far from optimal. The rest of this section
might be important. will address some of the most important remaining challenges
∙ Closed loop control - sensors and actuators are connected for a full scale deployment of IWSNs. In addition, we touch
to PID controllers that control the process with respect upon some research areas that are of less importance in process
to the actual set-point. The purpose of the control loop automation in short to medium term.
is to continuously stabilize the (instable) process by con-
trolling the actuators based on the actual sensor readings. A. Safety
Generally, closed loop control is sensitive to jitter and Safety of humans, environment, and property should always
delays. be the number one priority. In process automation some

412
functions are safety-critical by definition, but most of them overhead. The most problematic situation besides deliberate
are not safety-critical. This does not mean that only the safety- manipulation of control data is denial-of-service attacks, which
critical functions should be designed, developed, and certified cannot be prevented by cryptography at all. In case of a denial-
with care [12], as most process automation equipment depend of-service attack, the automation system will transition into a
on that the rest of the system operates within the boundaries safe state, if designed correctly.
of the specifications. Even if the functionality is not safety- Secondly, how to integrate the security mechanisms in the
critical by definition, there can be substantial production losses overall automation system in an efficient way with respect
or damages to the property if the automation equipment is to key management and replacement of faulty field devices
not designed to reduce the risk of uncontrolled or dangerous is an important issue. Users of control systems today are
situations. used to exchange faulty devices during operation without
The prevention of uncontrolled processes is extremely im- any additional configuration, i.e. “hot swap”, to minimize the
portant. As an example, if a set point from the control system downtime.
to a control valve cannot be transmitted, the valve should fall
back into a safe state (normally closed) after a time out. The C. Availability
time out depends on how long time the process can tolerate In industrial large-scale production availability is of sig-
a malfunction of the actuator before a possibly dangerous nificant importance. Even short and transient communication
situation occurs, ranging from milliseconds to seconds. In errors can cause significant production outages. This is mainly
addition to this, the control system should detect this com- due to that the process has to be stopped in a controlled manner
munication loss and indicate the failure. In this way, the rest in case of a single communication problem, and it can take
of the equipment that depend on the correct operation of the up to several hours to achieve full production rate again, with
control valve is signaled to avoid dangerous situations due to production losses in the range of hundreds of thousands dollars
error propagation. Non safety-critical automation equipment per hour. Self-healing mesh networks are appealing to use in
is designed such that if a problem can occur, it should be industrial automation for several reasons, i.e. for redundancy,
detected and force the process into a safe state. One of the availability, etc [14]. However, in literature it is commonly
worst scenarios that can occur is that the operator’s view in assumed that routing protocols in industrial settings should
the control room is not consistent with the actual state of the be able to deal with mesh networks containing thousands or
equipment on the factory floor. This implies that field devices tens of thousands nodes. Furthermore, it is assumed that all
cannot have extremely small duty cycles to preserve battery of the devices are battery operated, thus energy aware routing
since both the operators and control system will neither get protocols are of significant importance in order to distribute
any life sign from the field devices, nor meet the required the routing load in a fair manner amongst the nodes [4]. The
update frequency. first assumption is not correct, even though there are tens of
Some work with safety-critical communication using Wire- thousands of instruments that need to communicate, they do
less HART and PROFINET IO has been done in [13] and not belong to the same network because of the availability
the main result is that the IWSNs need to have deterministic concern. The nodes are distributed over a set of process
and synchronized communication in both the uplink and the controllers, divided in several process sections, in order to
downlink in order to avoid spurious fail-safe timeouts. avoid a complete production stop in case of for example
a non-fail-silent situation in one node. Furthermore, energy
B. Security optimized routing protocols might have a severe and negative
Most information that is transmitted to and from field impact on the latency and real-time performance of end-to-
devices is usually normalized valued of the measured entity, end communication using mesh networks in the case of link
i.e. 0-100% of the range of the measuring instrument. In problems caused by fading. In industrial automation the mesh
some cases the actual measurement is transmitted along with networks would rather benefit of rapid adoptions in routing,
the SI unit. The control system is collecting information of in the case of for example fading, while meeting the real-
the state of the process, and the process is controlled based time constraints. Flooding might be one feasible alternative
on the collected measurements and the control strategy by for usage in mission-critical wireless sensor networks [15].
transmitting set points for actuators based on the actual state
of the process. However, the main point is that it is not D. Latency/retransmission
confidential information as such that is transmitted. As the Due to the nature of automation the data transmitted in
confidential information resides inside the control system, i.e. the field networks is only valid for a short time. If the
recipes or control strategies, and those are not transmitted data is delivered too late it is of limited use, as in most
on the field networks. However, from a security perspective real-time systems. Therefore new data should be propagated
authentication, integrity, availability, and non-repudiation are through the network instead of guaranteeing delivery of all
important security objectives. With respect to this, the current transmissions. This is an important area of research, especially
situation is that confidentiality, authentication and integrity are within the area of IWSN where it can be mesh networks,
provided. Here optimizations can be made with respect to multi-hop situations, and synchronized communication in both
security to improve energy consumption, latency, and security directions between the nodes. In addition, the automation

413
systems will download configuration data to the field devices F. System Integration
both at startup and during operation that should be end-to- In order to make a smooth and efficient integration of
end acknowledged and retransmitted to be able to guarantee IWSNs into existing automation infrastructures, the most criti-
delivery in case of communication losses. To decrease the cal point to consider is the gateways. Today there exist a small
number of retransmissions in the IWSN one can use some error number of GW vendors and they are proposing proprietary
control techniques such as forward error correcting (FEC) solutions, which prevents an open and efficient integration to
codes [16]. In the commercial IEEE 802.15.4 based IWSN existing infrastructure. All commercially available solutions
today, FEC has been omitted mainly due power consumption today provide web-based configuration, everything is done
of decoding operation. Nevertheless, by using FEC the overall manually, which slows down the engineering, commissioning,
energy consumption will become less since we will spend less and maintenance of the system.
energy on retransmission [17] and re-scheduling our network. Today, most wired fieldbuses have services to download
The major trade off between the additional processing power configuration and to simplify the integration work in general.
and the associated coding gain need to be optimized in order What is currently missing is a standardized approach for IWSN
to have a power, energy-efficient and low-complexity FEC integration to the different fieldbuses in the IEC standards
schemes in industrial wireless sensor networks. In addition to [19]. Some standardization efforts are currently ongoing [20],
that, we need to consider memory constraints in the embedded [21] as well as proposals for integration [22]. Therefore it is
system and not jeopardizing the IWSN requirements given in essential that the services provided by the IWSN are possible
Table I. to integrate efficiently in the automation system to have a
With respect to retransmissions in mesh networks and multi- seamless transition from wired to wireless communication, as
hop situations, it has to be guaranteed that data is delivered well as simple deployment, commissioning, and maintenance.
in the correct order. It can easily occur that data is received
in the wrong order if packets are not discarded in the mesh G. Network Size
before the next periodic data is transmitted and delivered. The One important requirement for IWSN is scalability and
consequences can be substantial, as a single bit is used to start the networks capability to adapt to changing networks size.
and stop electric motors ranging from kilowatts to megawatts. Without this kind of support the network performance will
degrade significantly as the network size increases. In most re-
search work it is envisaged that the network (e.g., the gateway)
should be scalable and able to support 1000 sensor nodes. That
E. Lack of Support for Actuators
assumption is not valid in a process automation context where
it is more likely that the network will contain a maximum of 50
One great advantage with IWSNs is that it is not required to
nodes in order to meet the required refresh rates. This points to
route a communication cable from each sensor and actuator
that research needs to be focused on routing algorithms aimed
to the dedicated controller. Usually different sections of the
for smaller networks that are optimized for the requirements
process are distributed over several, sometimes redundant,
given in Table I, and where the key parameter is latency.
controllers to avoid that a failure in one part of the process
affects other process sections. The main reason is to limit H. Coexistence and Interference Avoidance
the consequences in case of a failure, and to increase the Industrial plants often contain many different wireless com-
availability of the plant by using buffers between the different munication technologies that operate in the same frequency
process sections. However, actuators would gain the same band that operate in the 2.4 GHz ISM band. Thus, it is very
advantage as for sensors with respect to this matter. Today, important that a solution can coexist in a radio environment
most standards lack support for actuators and therefore limits with a large amount of interferences as well as limit its own
this advantage, as a parallel wired infrastructure for actuators disturbance. The obvious countermeasure for interference is
has to be designed, deployed, and maintained. Of course, the use of different diversity (time, frequency, and space)
this have a major impact on the overall cost when deploying schemes [23] but also some more fancy innovative techniques
IWSNs today, and the cost benefit is partly eliminated due to such as interference cancellation, effective radio resource
this. management and software defined radios. WirelessHART [2]
Actuator support demands deterministic communication combines channel hopping (frequency diversity) and TDMA
from the automation system down to the actuators, and in but more sophisticated algorithms considering the combination
addition as described in Section IV-A the need for determinis- of multiple diversity schemes and scheduling of slots is needed
tic communication in both directions, as well as failsafe states in the future. It is indeed important to guarantee real-time
[18]. One of the main arguments against support for actuators services for IWSN [24] and therefore more research on real-
today, is that actuators need to be connected to the main grid as time scheduling for IWSN is of high priority.
a battery cannot provide sufficient energy to the actuator. What
you seldom read in literature is that many actuators actually are I. Energy Consumption
pneumatic. We will elaborate more on this matter in Section Research on energy consumption and energy harvesting
IV-I. / scavenging is a hot topic today. However, due to refresh

414
rates required in process automation it is difficult to save [2] (2010) Hart 7 specification. [Online]. Available: http://www.hartcomm.
much energy by minimizing the duty cycle of a wireless org/
[3] (2010) Isa 100, wireless systems for automation. [Online]. Available:
field device. Secondly, as previously mentioned actuators are http://www.isa.org/isa100
a mandatory part of process automation and they are some- [4] A. Willig, “Recent and emerging topics in wireless industrial com-
times pneumatic. Therefore, true wireless field devices are not munications: A selection,” IEEE Transactions onIndustrial Informatics,
vol. 4, no. 2, pp. 102 –124, May 2008.
foreseen in the near future in process automation. However, [5] I. Akyildiz and X. Wang, “A survey on wireless mesh networks,”
temporal installations of sensors used for validation of possible Communications Magazine, IEEE, vol. 43, no. 9, pp. S23 – S30, Sept.
process optimization will greatly benefit since they can be 2005.
[6] J. Yick, B. Mukherjee, and D. Ghosal, “Wireless sensor network survey,”
true wireless, operated on battery power. Future research in Computer Networks, vol. 52, no. 12, pp. 2292 – 2330, 2008.
energy scavenging will improve the situation and more kind [7] L. Zheng, “Industrial wireless sensor networks and standardizations:
of field devices can become true wireless. However, many The trend of wireless sensor networks for process automation,” in
Proceedings of SICE Annual Conference 2010, Aug. 2010, pp. 1187
devices are unlikely to become true wireless using today’s –1190.
technology, both sensors (i.e. acoustic or laser) and actuators [8] L. Hou and N. Bergmann, “System requirements for industrial wireless
(i.e. large valves or pneumatic control valves). Nevertheless, sensor networks,” in IEEE Conference on Emerging Technologies and
Factory Automation (ETFA’10), Sept. 2010, pp. 1 –8.
the advantages are still significant even if power supplies have [9] V. Gungor and G. Hancke, “Industrial wireless sensor networks: Chal-
to be wired to the sensors and actuators as the fieldbuses lenges, design principles, and technical approaches,” IEEE Transactions
are not necessary to route between all sensors, actuators, and on Industrial Electronics, vol. 56, no. 10, pp. 4258 –4265, Oct. 2009.
[10] J. Heo, J. Hong, and Y. Cho, “Earq: Energy aware routing for real-
controllers in the specific process sections. Power is almost time and reliable communication in wireless industrial sensor networks,”
always found close to the field devices today and is likely to IEEE Transactions on Industrial Informatics, vol. 5, no. 1, pp. 3 –11,
remain there, even in green field installations. 2009.
[11] M. De Biasi, C. Snickars, K. Landernas, and A. Isaksson, “Simulation of
Assume for a while that all field devices could be battery process control with wirelesshart networks subject to packet losses,” in
operated today, a new significant problem would occur, namely IEEE International Conference on Automation Science and Engineering
scheduled battery maintenance of tens of thousands of nodes in (CASE), Aug. 2008, pp. 548 –553.
[12] IEC 61508. Functional safety of electrical/electronic/programmable
a typical process industry. This would cause serious problems electronic safety-related systems. International Electrotechnical Com-
for the industries that are expected to operate flawlessly 24/7 mission, 2005.
with a few short schedule maintenance periods per year. In [13] J. Åkerberg, F. Reichenbach, and M. Björkman, “Enabling safety-
critical wireless communication using wirelesshart and profisafe,” in
addition, the end-users need to keep batteries of various size IEEE Conference on Emerging Technologies and Factory Automation
and capacity on stock in order to quickly replace exhausted (ETFA’10), Sept. 2010, pp. 1 –8.
batteries. That means in the end that the life-cycle cost for a [14] F. De Pellegrini, D. Miorandi, S. Vitturi, and A. Zanella, “On the use
of wireless networks at low level of factory automation systems,” IEEE
true wireless device would be more expensive compared to a Transactions on Industrial Informatics, vol. 2, no. 2, pp. 129 – 143,
device that only communicates wireless. May 2006.
[15] M. Soyturk and D. Altilar, “Reliable real-time data acquisition for
V. C ONCLUSION rapidly deployable mission-critical wireless sensor networks,” in IEEE
INFOCOM Workshops, 2008, pp. 1 –6.
Industrial wireless sensor networks (IWSNs) have been [16] L. Li, R. G. Maunder, B. M. Al-Hashimi, and L. Hanzo, “An energy-
efficient error correction scheme for ieee 802.15.4 wireless sensor
gaining acceptance during the last decade, largely due to the networks,” Transactions on Circuits and Systems II, vol. 57, no. 3, pp.
greatly increased flexibility, lower cost and scalability that they 233–237, March 2010.
have been shown to provide. Nevertheless, still there exist [17] M. Vuran and I. Akyildiz, “Error control in wireless sensor networks: A
cross-layer analysis,” IEEE/ACM Transactions on Networking, vol. 17,
several open areas in IWSNs that has not been addressed no. 4, pp. 1186 – 1199, Aug. 2009.
properly in existing research work. [18] J. Åkerberg, M. Gidlund, J. Neander, T. Lennvall, and M. Björkman,
In this article, we outline the most important research “Deterministic downlink transmission in wirelesshart networks enabling
wireless control applications,” in 36th Annual Conference on IEEE
issues that need to be solved immediately such that IWSNs Industrial Electronics Society (IECON’10), Nov. 2010, pp. 2120 –2125.
can meet the expected market requirements. Safety is by far [19] IEC 61784-1. Industrial Communication Networks - Profiles - Part 1:
the most important feature to consider and there are several Fieldbus profiles. International Electrotechnical Commission, 2007.
[20] (2010) Wireless cooperation team. [Online]. Available: http://www.
issues that need to be addressed before we can offer a safe hartcomm.org/hcf/news/pr2008/press conf interkama2008.pdf
and secure communication over wireless sensor and actuator [21] (2010) Tc2/wg12 wireless sensor/actor networks. [Online].
networks. For instance, it is important with downlink support Available: http://www.profibus.com/index.php?id=1314&tcwg tc uid=
2&tcwg wg uid=14
for actuators otherwise we cannot support safety applications. [22] J. Åkerberg, M. Gidlund, T. Lennvall, J. Neander, and M. Björkman,
Other important research topics are interoperability between “Integration of wirelesshart networks in distributed control systems
different wireless systems, availability and meeting real-time using profinet io,” in 8th IEEE International Conference on Industrial
Informatics (INDIN’10), July 2010, pp. 154 –159.
guarantees where latency is an important parameter to pay [23] D. Yang, M. Gidlund, and Y. Xu, “Coexistence of ieee 802.15.4 based
attention to. networks: A survey,” in 36th Annual Conference on IEEE Industrial
Electronics Society (IECON’10), Nov. 2010, pp. 2107 – 2113.
[24] S.-E. Yoo, P. K. Chong, D. Kim, Y. Doh, M.-L. Pham, E. Choi,
R EFERENCES and J. Huh, “Guaranteeing real-time services for industrial wireless
[1] T. Lennvall, S. Svensson, and F. Hekland, “A comparison of wirelesshart sensor networks with ieee 802.15.4,” IEEE Transactions on Industrial
and zigbee for industrial applications,” in IEEE International Workshop Electronics, vol. 57, no. 11, pp. 3868 –3876, Nov. 2010.
on Factory Communication Systems, May 2008, pp. 85–88.

415