Challenges and Conditions for Wireless Machine-to-Machine Communications in Industrial Environments

Transcription

1 TOPICS IN RADIO COMMUNICATIONS Challenges and Conditions for Wireless Machine-to-Machine Communications in Industrial Environments Peter Stenumgaard, Linköping University and Swedish Defence Research Agency (FOI) José Chilo, Javier Ferrer-Coll, and Per Ängskog, University of Gävle ABSTRACT Wireless solutions are rapidly growing in machine-to-machine communications in industrial environments. These environments provide challenging conditions in terms of radio wave propagation as well as electromagnetic interference. In this article, results from the characterization of radio channel properties are summarized in order to provide some guidelines for the choice of wireless solutions in industrial environments. In conclusion, it is essential to know the sensitivity of industrial processes to time delay in data transfer. Furthermore, it is important to be aware of the radio interference environment and the manner in which different wireless technologies react upon interference. These steps will minimize the risk of unforeseen expensive disturbances in industrial processes. INTRODUCTION Recent progress in wireless communication technology has resulted in a rapid increase in the use of wireless systems in industrial and factory environments. These applications consist of both voice communications between personnel as well as machine-to-machine (M2M) communications. In industrial environments, wireless communication links often transfer vital information between machinery, control, and monitoring devices. In critical applications, this information has one essential requirement: timely delivery without failure. However, standard wireless technologies (e.g. WLAN and Bluetooth) are usually designed for the office or outdoor environment where the building structure is different from the structure seen in many industrial environments [ 3]. The electromagnetic interference in office environments also differs significantly from the electromagnetic interference measured in industrial environments. Consequently, the conditions for communications are completely different. Furthermore, such technologies rely on retransmission mechanisms such as carrier sense multiple access (CSMA) or automatic repeat request (ARQ), which lead to increased delay if radio interference is present. This can be a challenging issue because many industrial environments exhibit high levels of electromagnetic interference. Additionally, the physical contents of storage and production halls can play an important role in wave propagation. Our research shows that in certain industrial environments such as a paper warehouse in a paper mill, there are highly absorbing environments where wave propagation is characterized by single-path propagation with high attenuation. Therefore, a thorough understanding and description of several radio interference properties for these environments is important for the choice of wireless technologies and the development of new or improved wireless standards for industrial applications. These environments can be characterized using well understood parameters such as power delay profile, temporal dispersion properties, and propagation path loss. Having such results, convenient technology solutions for M2M applications can be chosen for a specific class of industrial environment. By choosing convenient technology solutions for such applications, the risk of expensive interference problems causing production disturbances can be reduced. Furthermore, the benefit from technology improvements such as adaptive interference management [] can be well analyzed for a specific class of environments. The article is organized as follows. In the next section, some general findings about electromagnetic interference in industrial environments are summarized. This is followed by a summary of the variety with respect to multipath propagation for those industrial environments we have investigated. The challenges connected to time-critical communications are analyzed, and an overall comparison of time-delay properties IEEE Communications Magazine June /3/$ IEEE 87

2 Amplitude (dbm) Spectrum Frequency (MHz) Figure. Electromagnetic interference measured in a boiler house. between some typical wireless standards is shown. Advantages and drawbacks of industrial, scientific, and medical (ISM) frequency bands are also discussed. Some general findings about convenient technology solutions for increased robustness of wireless solutions are given for different generic industrial environments. Finally, the article is concluded. ELECTROMAGNETIC INTERFERENCE Typical sources of high levels of radiated electromagnetic interference in industrial environments are: Electrical engines Power converters Charging devices for battery-driven equipment Frequency converters Welding processes using pulsed power Other wireless systems Personal computers Welding due to repair and maintenance work can cause unexpected interference problems since interference signals from pulsedpower welding can have a frequency content up to several hundreds of megahertz. Furthermore, such work can be difficult to foresee since the need can arise rapidly and without recognition of possible interference risks with wireless solutions in a production hall. The electromagnetic interference from modern PCs typically occurs for frequencies from some tens of megahertz up to 2 3 GHz. Here, interference incidents have been seen between PCs and wireless headphones intended for production halls with noisy environments. Other typical interference incidents are with wireless door openers, remote-controlled cranes, and wireless sensors for control and monitoring of processes. These kinds of incidents normally cost lots of money due to disturbances of production lines. A few measurement results have been published for different individual electric machines [8, 9]. No measurements of the total interference levels in complex industrial production halls have been found in the literature. Thus, we have performed such interference characterization measurements for different industrial environments. In general, the typical electromagnetic interference in industrial environments has the properties exemplified by Fig.. The interference is measured in a boiler house where numerous electrical engines, power converters, and frequency converters are used. For lower frequencies, the level of interference increases. In Fig., it can be observed that the increase in level is more rapid when the frequency is decreased from a breakpoint in frequency of approximately 5 MHz. The frequency value for this breakpoint changes with the specific environment, but is typically between 2 and 5 MHz for the environments studied. The breakpoint is important to identify since several wireless solutions for industrial applications use frequencies in the 3 8 MHz and 4 45 MHz bands. The peaks around 9 MHz and 8 MHz are from GSM signals in the area. The peaks at MHz come from a DECT system in the area. MULTIPATH PROPAGATION The characterization of industrial environments requires information about the radio channel. The power delay profile (PDP) plays a key role, and depends mainly on the field strengths and delay times of incoherently impinging reflected waves. Root mean square (RMS) delay spread and maximum excess delays are typical parameters that describe the time dispersion in the channel. The RMS delay spread is a measure of how long the impulse response of the radio channel is on the average. The maximum excess delay provides information about the time at which delayed components can arrive at the receiver after the first radio wave component. Characterizations of multipath propagation in indoor industrial environments have been published for a few cases for higher frequencies [6, 7]. However, several industrial wireless applications also use lower frequencies such as the 4 MHz frequency band. Thus, we have also done characterization measurements for this part of the frequency band in our investigations. Figure 2 shows examples of impulse responses for two industrial environments having fundamental differences with respect to multipath propagation. The highly reflective environment is typically found in a hall with metallic structures covering the floor, the walls, and the roof. The highly absorbent environment is typically seen in a storage hall where electromagnetic absorbent material is stored. In our study, a highly absorbent environment is represented by a large storage hall for finished paper products. In a highly reflective environment, the number of multipath components is much greater than the number of multipath components in highly absorbent environments. Furthermore, the delay these components experience is higher in the case of highly reflective channels. One important consequence of these differences is that multi-antenna technologies will provide large improvements in the highly reflective 88 IEEE Communications Magazine June 23

3 .8 RMS delay = ns Max excess delay = 898 ns.8 RMS delay = 2.6 ns Max excess delay = 2 ns RMS delay = ns Max excess delay =2 ns.8 RMS delay = 23.5 ns Max excess delay = 8 ns RMS delay = ns Max excess delay = 86 ns.8 RMS delay = 28.9 ns Max excess delay = 42 ns Figure 2. Power delay profiles for highly reflective and highly absorbing environments at 433 MHz, 89 MHz, and 245 MHz bands for non-line-of-site (NLoS), top to bottom. environment, but almost no improvement at all in the highly absorbent environment. On the other hand, in the highly absorbent environment, a single-path channel will allow a very simple receiver and high symbol rates. Figure 3 presents the cumulative distribution function (CDF) of the delay for highly absorbing, office, and highly reflective environments, and shows a plot of the total received energy vs. the delay experienced by the multipath components. The measurements for the office environment were collected in typical offices at the Radio Centre in the University of Gävle. When the total received energy is 9 percent, the delays measured for a typical office environment, a highly absorbing environment, and a highly reflective environment are 48 ns, ns, and 664 ns, respectively. We observe large differences in the number of components and the maximum excess delay. In the case of highly reflective environments, some components arrive more than 2 ns after the first component, which can cause problems in systems with a high symbol rate. In the highly absorbing environment, the maximum excess delay is 3 ns, which is significantly lower than in the case of highly reflective environments. In the highly reflective environment, 23 components were captured, while in the highly absorbent environment, 27 were captured. Figure 4 represents an approximate combined summary of the findings from both the interference and multipath measurements for different environments. The maximum excess eelay and the typical received interference power for a typical radio receiver are shown along the X-axis and Y-axis, respectively. Figure 4 depicts the scenario that can be expected in different industrial environments. The newly designed boiler house has both higher interference levels and IEEE Communications Magazine June 23 89

4 Amount of total received power longer impulse responses. The interference level is due to more electronic solutions being used such as electro filters for smoke cleaning, and a larger amount of electrical engines and frequency converters to manage the more advanced process. As seen, the interference level at railyards increases if the electric cargo trains carry heavier loads. Furthermore, the general observation from our measurements is that steel production environments exhibit moderate interference levels as well as moderate length of the channel impulse response. High absorbing Office environment High reflective Delay (s) x -6 Figure 3. Cumulative distribution function in highly absorbing, office, and highly reflective environments, NLoS case. TIME DELAY FOR CRITICAL APPLICATIONS For the choice of wireless solutions in industrial applications having stringent maximum time delay requirements, it is crucial to know about the interference environment and the manner in which it affects different communication standards. Several common standards result in increased time delay when the amount of interference increases. In order to choose a wireless technology for a critical industrial application, it is important to know the manner in which different wireless solutions react to radio interference. In general, the first step is to determine whether a wireless solution developed for the ISM frequency band or a solution for a dedicated industrial frequency band should be used. Three common ISM bands are 433 MHz, 868 MHz, and 2.4 GHz. ISM bands tend to be popular since several cheap solutions are available in the market, and no special permission is needed to use them if the equipment fulfills certain fundamental requirements. The drawback with ISM bands is the radio interference from other users caused by the widespread usage of these licensefree bands. For example, in the band GHz, WLAN (IEEE 82.x), Bluetooth (IEEE 82.5.), and ZigBee (IEEE ) exist. These technologies react with increasing data delay if interference occurs. A schematic comparison between some short-range technologies is shown in Fig. 5. Data for technologies marked with *) is obtained from [4]. For comparison, some approximate values from mobile phone standards are shown. The comparisons performed in Fig. 5 are always debatable since the exact values are dependent on the basic assumptions made. However, in our study, this comparison is sufficient in order to obtain some basic knowledge of the differences between these standards with respect to delay when interference occurs. It is seen that WLAN is a technology that can produce high data rates if no interference occurs, but can result in very high delays when interference is present. ZigBee is a technology developed with the objective of creating large mesh networks with low energy consumption for monitoring purposes. WISA is a technology developed by ABB, and is adapted for real-time monitoring and remote control of time-critical industrial applications. For reference, some approximate maximum time delays for mobile standards such as GSM, third generation (3G), and 4G are shown. Interference level (dbm) Cargo train high load Mine tunnel train Large industrial halls Railyard in general Rolling mill Paper production Storage hall steel sheets Paper store Maximum excess delay (ns) Newly designed boiler house Old boiler house Figure 4. A summary of the interference levels and maximum excess delay for the environments investigated. TECHNOLOGY SOLUTIONS FOR INCREASED ROBUSTNESS From our research findings, we can propose technology solutions that provide the largest improvements with respect to increased robustness and reliability for wireless solutions in industrial applications. In Table, some improvement measures have been proposed for four generic environment classes that represent most industrial environments. The large variety of conditions for wireless communications in the environments investigated leads to the conclusion that the most convenient technology solution for a specific environment differs. By considering the characteristics for this environment together with the specific requirements for a certain wireless service, convenient technology solutions can be chosen. 9 IEEE Communications Magazine June 23

5 For example, if we are operating in an environment with high interference levels and have requirements for low time delay, we should avoid technologies based on retransmission to handle interference since these will cause time delays. An example is WLAN, which uses CSMA that listens to the channel before transmitting and waits until a free channel is observed. For such a link, Bluetooth could be a possible choice if the requirement on available data rates is moderate. WLAN is a convenient choice if we wish to transmit high data rates in an environment with low levels of interference and moderate requirements for time delay. If we wish to build a mesh network in an environment with low interference levels, ZigBee could be a possible choice. However, if the interference level is high, WISA technology will be a better solution. CONCLUSION Industrial environments exhibit significantly different behavior for wireless communications than do indoor office environments or outdoor environments. The interference levels are generally higher, and the multipath propagation is generally more complex because of the large number of metallic surfaces. There are also large varieties in interference levels and multipath propagation between locations within a single industrial plant. When wireless solutions are used for time-critical applications, it is of great importance to know the manner in which different industrial processes react to time delay in data transfer. Furthermore, it is essential to be aware of the radio interference environment and the manner in which different wireless technologies react to interference. If these steps are followed, the risk of unforeseen expensive disturbances in industrial processes is minimized. In this study, we have summarized our findings for characterization of industrial environments with respect to the conditions for wireless transmission. These findings can be used as approximate guidelines for the choice of wireless solutions in critical industrial applications. REFERENCES [] J. Chilo et al., EMI Disruptive Effect on Wireless Industrial Communication Systems in a Paper Plant, Proc. IEEE EMC Symp. Austin-Texas, vol. 3, Aug. 29, pp [2] J. Chilo et al., Characterizing Electromagnetic Interference in Vicinity to a Railway Freight Train, Proc. IEEE Int l. Symp. EMC/EMECO, Saint Petersburg, Russia, June 29, pp [3] J. Chilo et al., APD Measurements for Characterization and Evaluation of Radio Interference in Steel Mill, Proc. IEEE Int l. Symp. EMC, Kyoto, Japan, vol. 2, July 29, pp [4] G. Scheible et al., Unplugged but Connected: Design and Implementation of a Truly Wireless Real-Time Sensor/Actuator Interface, IEEE Industrial Electronics Mag., vol., issue 2, 27, pp [5] J. F.-Coll et al., Characterization of Electromagnetic Properties in Iron-Mine Production Tunnels, IET Electronics Letters, vol. 48, no. 2, Jan. 22 [6] J. Kåredal et al., UWB Channel Measurements in an Industrial Environment, IEEE GLOBECOM, vol. 6, Nov. 24, pp (ms) Low delay High delay. Low data rate Figure 5. A comparison of different wireless standards in terms of time delay when electromagnetic interference is present. Generic environment Highly reflective Highly absorbing ENOCEAN*) GSM (EDGE)) Medium reflective Tunnels for industrial use IEEE *) (ZigBee, W-HART) IEEE 82.5.*) (Bluetooth) WISA*) [7] M. Sánchez Varela and M. García Sánchez, RMS Delay and Coherence Bandwidth Measurements in Indoor Radio Channels in the UHF Band, IEEE Trans. Vehic. Tech., vol. 5, no. 2, Mar. 2, pp [8] J. Catrysse, J. Rayée, and D. Degrendele, Study and Simulation of the Ambient Noise of an Industrial Environment for Wireless Communication Applications, EMC Europe Wksp. 26, Rome, Italy. [9] J. Catrysse et al., In Situ Testing of Large Machines: Alternative Method for Radiated Emission Measurement, IEEE EMC 28 Int l. Symp. Electromagnetic Compatibility, 8 22 Aug. 28, pp. 6. [] Villa et al., Adaptive Modulation and Coding with Hybrid-ARQ for Latency-Constrained Networks, 8th European Wireless Conf., 8 2 Apr. 22 pp. 8. BIOGRAPHIES High data rate IEEE 82. x*) (W-LAN) 3G (HSPA) PETER STENUMGAARD (peter.stenumgaard@foi.se) is an adjunct professor at the Division for Communication Systems in the Department of Electrical Engineering (ISY) at Linköping University (LiU), Sweden. He joined LiU in March 2. He received his Ph.D. degree in radio communications from the Royal Institute of Technology, Stockholm, in 2. Until 995 he was a systems development engineer at Saab Military Aircraft, Linköping, where he worked with electromagnetic compatibility issues in aircraft design. He is currently the research director in robust wireless communications at the Swedish Defense Research Agency (FOI) in Linköping. During he was an adjunct professor at the Center for Radio Measurement Technology at the University of Gävle, Sweden. He is also the director of the 4G (Mb/s) Voice: < 5 2 ms Technology providing the greatest improvement Multiple antenna solutions (diversity, MIMO) Equalizing Frequency selection adapted to the frequency dependence of the wave propagation properties Multiple antenna solutions Frequency selection adapted to the frequency dependence of the wave propagation properties Frequency selection adapted to the frequency dependence of the wave propagation properties [5] Equalizing. Maximum excessive time delay can be estimated from the geometry of the tunnel. Table. A summary of measures for increased robustness of wireless systems in different generic industrial environments. IEEE Communications Magazine June 23 9

6 graduate school Forum Securitatis for public safety and security at Linköping University. JOSÉ CHILO received his B.Sc. degree in industrial engineering from the University of San Agustin Arequipa, Perú in 986, and his B.Sc. degree in electrical engineering from KTH in 2. From 986 until he began his studies at KTH in 998, he worked as a consultant in various industries such as textiles, garment manufacturing, and transportation, in both Perú and Sweden. He received his Ph.D. in physics from KTH in 28, where he conducted research on signal and data processing, in particular advanced measuring techniques and event classification. He is presently an associate professor and senior lecturer at the University of Gävle in Sweden. JAVIER FERRER-COLL received his degree in telecommunication engineering in 28 from the Universidad Politécnica de Valencia. He is currently a Ph.D. student in the School of Information and Communication, KTH, and working at the University of Gävle. In 22, he received his Licentiate of telecommunications engineering from KTH. His main interest is channel characterization in industrial environments, especially interference detection and suppression. PER ÄNGSKOG graduated as an electronics engineer at the University of Gävle in 987. Between 987 and 2 he worked in Ericsson AB. Until 99 he conducted pre-studies of Digital Radio Frequency Memories (DRFM) for airborne electronic countermeasures. From 99 to 2 he worked with design of measurement systems for testing of transmitters and receivers aimed for mobile telephony systems. Since 2 he has been employed by the University of Gävle, where he currently teaches in the Master s program in telecommunications and occasionally gives courses in RF to external companies. Since 27 he has also been a research engineer in research projects at the Center for RF Measurement Technology, University of Gävle. 92 IEEE Communications Magazine June 23