Performance Evaluation and Prediction of a Bluetooth Based Real-Time Sensor Actuator System in Harsh Industrial Environments

Transcription

1 Performance Evaluation and Prediction of a Based Real-Time Sensor Actuator System in Harsh Industrial Environments U. Meier, S. Witte, K. Helmig Institute Industrial IT University of Applied Sciences Lippe Höxter Lemgo, Germany uwe.meier@fh-luh.de M. Höing, M. Schnückel, H. Krause Weidmüller Interface GmbH & Co. KG Klingenbergstraße Detmold, Germany Michael.Hoeing@weidmueller.de Abstract This paper deals with a performance evaluation of a based industrial real-time sensor actuator interface for stationary and mobile applications. The investigations were carried out in harsh industrial environments in the presence of parasitic machine emissions, intentional and unintentional interferers. The air paths were obstructed by moving machine parts and daily work traffic. As a result we can conclude, that wireless automation systems based on technology are extremely reliable due to their inherent system features like adaptive frequency hopping at high operating frequencies, error detection and correction. Neither parasitic machine emissions nor other radio systems or transceiver movements can impair the transmission, as long as distances are below 30 m. This contribution looks at an industrial sensor actuator interface () based on technology and provides information about the limits of interference resistance and transmission reliability of process data transfer in industrial fields. It shows the environmental impact on the primary physical parameters as well as the consequences of these impacts to a Profibus system. Performance prediction is provided based on measurements and appropriate simulation models. 2. Wireless Sensor Actuator System s are field devices for reading in sensor signals and driving actuators. In our system up to six wireless s can be linked to one gateway which is connected to a via a wired Profibus field bus (Fig.1). 1. Introduction Numerous innovative applications of automation technology require the usage of wireless technologies even for transmission of automation signals in harsh IP67 environments. To avoid later disappointments technological limits should be considered early in the initial planning stage. A given wireless technology is not appropriate for all applications: a stationary data link in a large machine production hall demands other wireless requirements than a local robot assembling cell, where the mobile robot unit moves with velocities up to 60 km/h. In addition to these passive impairments active sources of disturbances have to be considered as well. These are parasitic machine emissions and unintentional or even intentional other wireless systems. Numerous papers consider the impact of interferers on the performance of WLAN [1-3], but only little research has been down to study the performance itself, especially in harsh industrial environments [4-6]. Fieldbus () - Figure 1: Up to 6 s can be linked to one gateway Both, the module and the gateway, are realized with a robust housing meeting IP67 requirements (Fig.2). For these requirements an integrated antenna with an almost isotropic radiation pattern was designed.

2 Figure 2: based module with integrated antenna Fig.4 shows the measurement set-up for delay measurements. A function generator serves as sensor emulator. It generates input signals at the wireless s and at the gateway for comparison purposes. The s return this sensor information polling based via the bluetooth link. A Profibus line connects the gateway to the, where the input signals are available. As the function generator triggers the oscilloscope as well, this setup reveals important information on the overall system response time with respect to external events. The data transmission uses the WIRELESSopen protocol [7], which is similar to CANopen, together with the serial port profile SPP. All presented measurement and simulation results use the asynchronous transmission mode ACL with its shortest packet length type DM1. Forward error correction (FEC) and automatic repeat request (ARQ) with maximal 5 packet repetitions were used for the measurements. oscilloscope function generator s 3. System Parameters and Measurements Important system parameters at the application layer are packet losses (PL), bit error rate (BER), and delay (DEL). The delay can be further specified with minimal delay, maximal delay, and jitter. The latter is the difference between minimal and maximal delay. Analyzer Figure 4: Measurement set-up for delay measurements 4. Industrial Environments Several industrial environments (IND) were investigated in addition to some university labs (LAB): IND 1: Mechanical production hall, 12 m distance, line of sight (LOS), moving persons IND 2: Mechanical production hall, 12 m distance, obstructed line of sight (OLOS), moving persons IND 3: Mechanical production hall, 3 m distance, OLOS, slow machine movements Figure 3: Measurement set-up with protocol analyzer for PL and BER measurements Fig.3 shows the measurement set-up to determine PL and BER. A traffic application was realized where the requests the via a profibus gateway cyclically. At each request the responds a return message of 7 byte: 1 byte identifies the. 1 byte serves as polling byte which is incremented after each request. 2 bytes count the order of the messages. This counter is incremented after each request. A wrong order detects packet losses. 2 bytes are used for bit error detection (BER). 1 byte returns the link quality indication (LQI) as determined by the. This figure is related to the BER at the baseband layer as seen by the. Figure 5: Assembling machine (IND 4). Yellow circle shows location of.

3 IND 4: Connector assembling machine, 4 m distance, LOS, fast cyclic machine movements (Fig. 5) IND 5: Mechanical production hall, 3 m distance, LOS, sparc erosion machines were operating IND 6, IND 7: warehouse with high shelf cantilever system, NLOS, distance range m, cantilever speed up to 6 km/h, was located in a plastic box IND 8: robot based production cell, LOS, moving with 13 km/h, distance range m 5. Measurement Results A spectrum investigation in different industrial environments showed no substantial impairments of the used 2.4 GHz band. Even the worst parasitic emission of a sparc erosion machine showed only spurious emissions up to 1.5 GHz. Thus, only other radio systems in the same frequency band need to be considered. Necessary for a possible substantial interaction of different radio systems are interference power levels of at least the same order. We carried out several tests with a signal-to-interference ratio of SIR = 0 db. The worst interferer for is WLAN, which covers a frequency band of 20 MHz, i.e. 20 channels. In addition to interfering effects we carried out several measurements with stationary and mobile s. Fig.6 shows the RF transmission factor, i.e. the channel gain, of a mobile robot application. It was measured with a spectrum analyzer. Depending on the position, channel losses between 55 db and 78 db may occur. Even more channel losses up to 107 db were measured in the warehouse environments IND6, IND Tab.1 shows results acc. to the measurement set-up in Fig.3. The WLAN interferers were located in close vicinity to the measurement link in order to achieve SIR = 0 db at the location. No bit errors were detected, i.e. BER = 0. This is most important for the prospective application. Packet losses were noticed only in selected cases of Tab.1. The highest PL figures of IND6 and IND7 were caused by the large amount of channel loss. Distances of more than 50 m and no line of sight indicate a violation of possible application limits. Without additional strong WLAN interferer, we can conclude error free and reliable transmission in almost all industrial environments. Table 1: Packet losses were detected only in these environments. WLAN interferer with SIR = 0 db. environment LAB, 3m, 2 interferers LAB, 12m, 1 interferer LAB, 12m, 2 interferers measurement erroneous relative cycles cycles errors 21, , , IND 4 9, IND 6 1, IND 7 2, Results of the delay measurements can be seen in Fig.7 and Tab.2. The gateway response comes first as it is directly connected to the. The wireless s are polled at intervals between ms by the gateway. This polling time determines the jitter of the system. The jitter is always larger than the polling time. external signal time / sec gateway 2 Figure 6: Time varying channel gain between gateway and for the robot application IND 8. Figure 7: Delay measurements acc. to Fig.4

4 Table 2: Results of delay measurements No. of s polling time minimal delay jitter 1 20 ms 10 ms 32 ms 2 20 ms 10 ms 33 ms 3 25 ms 10 ms 48 ms 6. Evaluation and Performance Prediction Pathloss is the most important parameter for optimal system operation. Fig.8 shows two linear increasing prediction curves for ideal free space (blue line) and a laboratory environment (green line). They are based on the equation L/dB = n log 10 (d/m) with path loss exponents n = 2 for ideal free space and n = 3 for lab environment. All measured path loss values were found between these two prediction curves. path loss / db 100,00 80,00 60,00 40,00 Ideal Free Space AC LAB min LAB max IND min IND max Li it LAB 20,00 1,00 10,00 100,00 distance / m Figure 8: Average pathloss of different environments in the 2.4 GHz band with 0 dbi antennas. Measurements: anechoic chamber (AC), university lab (LAB), industrial environments (IND) Given a maximal transmitter power of 20 dbm (100mW) and a receiver sensitivity of 80 dbm (10 pw) we can derive a maximal pathloss of 100 db. Taking into account frequency selective fading with pathloss variations of ±15 db, the maximal average pathloss should not exceed 85 db. From Fig.8 we can derive a maximal distance of m for a reliable system performance. Another important parameter to control the system performance is the number of allowed ARQ packet repetitions n ARQ. It defines the error correction performance and the real-time requirements. Fig.9 shows simulated results of packet losses PL SPP and bit errors BER SPP at the application layer SPP. The simulation was performed with Matlab/Simulink. It takes into account the data packet handling as specified in IEEE , like packetizing, forward error correction, CRC check, and GAUSSIAN binary FSK modulation including frequency hopping [10]. The industrial channel models were AWGN channels with noise power of 114 dbm and 51 db path loss for the 3 m distance and 40 db path loss for the 12 m distance with RICIAN fading: K = 5, delay vector [0, 30 ns, 80 ns], gain vector [0, 1.5 db, 3 db]. in % 2,5 2 1,5 1 0, n_arq Figure 9: Simulated results of varying packet losses PL and bit error rate BER at the SPP layer as a function of maximal packet repetition n_arq in industrial environments As can be seen from Fig.9 transmission without ARQ should be avoided if a low rate of transmission error is desired. Even one allowed repetition can eliminate bit errors on the SPP layer. Avoiding packet losses can require up to five packet repetitions. 7. Summary Wireless automation systems based on technology are extremely reliable due to the inherent system features like adaptive frequency hopping at high operating frequencies, error detection and correction. Neither parasitic machine emissions nor other radio systems or movements can impair the transmission as long as distances are below 30 m and non-directional antennas are used. Larger distances are possible but require directional antennas. References PL_SPP (3 m) BER_SPP (3 m) PL_SPP (12m) BER_SPP (12 m) [1] A. Doufexi, et al.: An Investigation of the Impact of Interference on the Performance of g Wireless Local Area Networks, IEEE 2003, pp

5 [2] K. K. Wong, T. O'Farrel: Coverage of g WLANs in the Presence of Interference, IEEE 2003, pp [3] J. Park, et al.: Effect of Interference on OFDMbased WLAN, IEEE 2003, pp [4] U. Bilstrup, P.-A. Wiberg.: in Industrial Environment, IEEE WFCS-2000, pp [5] Y. C. Fai, et al.: enabled mobile robot, IEEE Int. Conf. on Industrial Technology 2002, pp [6] N. Golmie.: Interference in the 2.4 GHz Band: Impact on the Access Control Performance, NIST 99 [7] M. Schnückel, S. Witte: Keine neuen Drähte - Drahtlose Kommunikation von Automatisierungskomponenten, ELEKTRONIK 1/2005, S [8] M. Höing, K. Helmig, U. Meier: Erprobungstests von drahtlosen Sensor-Aktor-Systemen in rauen Industrieumgebungen, SPS/IPC/DRIVES, November 2006, Nürnberg [9] M. Höing, K. Helmig, U. Meier: ungestört - Erprobungstests der Technologie am Beispiel eines industriellen Sensor-Aktor-Systems, Wireless Automation 2007, Magdeburg [10] A. Bhardwaj: Investigation of Wireless Channel Models for Industrial Applications, Master thesis, UAS Lippe Hoexter, 2006 Acknowledgement The authors are grateful to Mr. T. Ginzel of OWITA GmbH, Lemgo and Mr. J. S. Michels of Weidmüller Interface GmbH & Co, Detmold for valuable discussions. This project was funded by the German ministry of education and research BMBF, project number 1769X05.