Physical Layer for Industrial Radio Fieldbus Networks

Transcription

1 Physical Layer for Industrial Radio Fieldbus Networks C. Koulamas (1), A. Lekkas (1), G. Kalivas (2), S. Koubias (2), G. Papadopoulos (1) (1) Industrial Systems Institute (I.S.I.) (2) Dept. of Electrical and Computer Engineering University of Patras Patras, Greece Tel: , ABSTRACT In this paper, a radio physical layer for a wireless fieldbus network is proposed. For this, a model is described and a performance evaluation is presented, by means of both measurements and analytical calculations of the Bit Error Rate (BER), using the Direct Sequence Spread Spectrum (DSSS) technology in static and dynamic modes, and of the timing of the basic message cycle, using Profibus as the fieldbus MAC. I. INTRODUCTION The benefits of a potential usage of wireless communications inside the industrial environment are well recognized. Nevertheless, a number of dependability and performance aspects, still defer the wide adoption of wireless technologies inside the real automation environment, in contrast to the degree of acceptance of standard, wired fieldbus technologies, which support numerous hard real-time industrial applications of today. The work presented in this paper is done in the framework of the IST project R-Fieldbus. The main objective of R-Fieldbus is to design and deploy an innovative high-performance radio fieldbus (R- FIELDBUS) capable to cope with harsh industrial environments and support both traditional fieldbus realtime control services and extended application services like industrial multimedia. The objective of this paper is the validation of the proposed radio physical layer against the operational requirements in industrial type of environments. The paper is structured in four main sections. Section II presents a number of basic requirements for a wireless fieldbus, as these were derived from a user survey [1] accomplished in the context of the R-Fieldbus project, summarizing also the results of the technology assessment made in the same context. Section III describes the basic reference model proposed for the physical layer of an industrial radio fieldbus network. Section IV validates the DSSS radio technology by presenting propagation and BER measurements in an industrial environment, while section V exhibits the performance of the physical layer components in the time domain. This work was supported by the IST RFieldbus project II. RADIO FIELDBUS REQUIREMENTS & SOLUTION ALTERNATIVES Any successful wireless fieldbus application should conform to two kinds of requirements, which are the radio technology requirements imposed by the wireless nature of communication and the industrial applications and user requirements, imposed by the current industrial practice and fieldbus technology [3]. The radio technology requirements focus on the inherent ability of the system to communicate satisfactorily over the radio channel. Taking into account the harsh industrial environment, these requirements are summarized in Table 1. Bit Rate 1-2 Mbit/s BER < 10-5 Range 100 m Path Loss Delay Spread ~100dB > 200 nsec Table 1:Radio Fieldbus requirements Bit-rate and range could be also categorized among the user or application related requirements, the rest of which are: Fieldbus equivalency: the real-time and the dependability quality of the wireless extensions shall be comparable with these of the currently used wired solutions. Fieldbus compatibility: the success of any wireless technology inside industry is expected to be reached if this technology integrates seamlessly with the current industrial networking infrastructure, being capable to provide an extension to it. Detailed assessment of the most popular radio solutions of today [2], [3], show that a number of these technologies must be ruled out due to either their inability to operate satisfactorily under severe multipath (DECT, Bluetooth), or to the lack of support of the required bit-rate (DECT, Bluetooth) or a combination of requirements (802.11b-FHSS, UMTS-FDD). From the remaining existing solutions, strong candidates appear to be the b-DSSS, the Hiperlan-2 and the UMTS-TDD wireless protocols. Timing analysis of the full-stack deployment of these protocols [3], [4], show that the introduced delays are far from being comparable to any existing wired fieldbus solution, leading to a contradiction to the

2 16,04 m fieldbus equivalency requirement stated above. Moreover, the above mentioned fieldbus equivalency and compatibility requirements should make evident that the final fieldbus protocol stack must be preserved down to the layer which provides the real-time services, that is, the Data Link Layer. This would lead to a redundancy, if we preserved the full protocol stack of the existing wireless protocols, and thus, to a performance degradation and a complexity increase. Therefore, we focus in the adoption of only the physical layers - as they are specified in the standards of the wireless protocols mentioned above inside a standard wired fieldbus protocol stack. Regulations concerning the UMTS technology prohibit this and require always its full-stack adoption. Therefore, the final alternatives are to use the Hiperlan-2 OFDM physical layer burst, or the b DSSS physical layer burst under a standard fieldbus MAC. Due to the lack of wide availability of OFDM products, the rest of the paper is focused on the performance of the DSSS alternative, but the architecture of the model and the definition of the physical layer service make possible the seamless adoption of any other radio transmission technology, which satisfies the requirements. III. PHYSICAL LAYER MODEL The main objective of the radio physical layer model is to provide for both the seamless exchangeability of any part, component or sublayer with a different existing or upcoming technology (i.e. DSSS vs OFDM modem bursts) and for the seamless adoptability of the whole layer in an existing fieldbus protocol stack (i.e. IEC-61158, EN-50170). PhL-User entity PhL Interface PhL DIS (DCE Independent Sublayer ) DIS - DCE Interface PhL DCE (Data Communication Equipment) PhL-User Management Entity PhLME Interface PhLME (PhL Management Entity) PhLME Interface Station Management completed by the addition of the Layer Management Entity (LME), which hides the management of all the parameters for the operation of the physical layer. IV. DSSS DCE (MODEM) PERFORMANCE A. Description of the site The site selected for the measurements is located at the University of Magdeburg [2]. It is a manufacturinglike environment and it is representative of a heavy industrial environment. The dimensions of the site are 16mx45m and 10m high. Figure 2 depicts the measurement site. The building is composed by three levels, one open basement and two open metallic scaffolds. Since it is part of the machine-building faculty of the University, there are road construction machines, cranes, metallic shelves and workbenches and other mechanical structures. Its external walls are made of concrete while there are large glass windows. The whole site includes the various types of environment known and classified in radio propagation. These are: Line of Sight with Light surrounding clutter (LOS-LC), Line of Sight with Heavy surrounding clutter (LOS- HC), Obstructed Path with Light surrounding clutter (OP-LC) and Obstructed Path with Heavy surrounding clutter (OP-HC). Figure 2 depicts the ground floor plan of the site and shows the mapping of the positions selected for the measurements. Rectangles and parallelograms indicate position of the metal structures in the hall. Twenty-four positions are selected in all three floors, so as to encompass all the propagation scenarios and represent most possible layout scenarios in the deployment of industrial radio communication system. The measurement campaign is divided into two sections. First propagation measurements are performed to identify the most crucial radio channel parameters such as path loss, coherence bandwidth, average and RMS delay spread [9], [10]. In the second section, Bit Error Rate (BER) and Packet Error Rate (PER) measurements are performed in order to investigate the PHY layer performance of DSSS modems. BER and PER are examined as affected by: received power, delay spread and channel profile, antenna diversity, antenna directivity and adjacent channel interference. Figure 1 Radio Fieldbus PhL Model 3 m 5 m 5 m The proposed physical layer model for a radio fieldbus network is presented in Figure 1. The model is mainly influenced by the physical layer specifications in both the IEC [5] (fieldbuses) and IEEE [6] (wireless networks) standards. It consists of two main sub-layers, namely the Data Communication Equipment (DCE) sublayer, which hides the selected modem techniques, and the DCE Independent Sublayer (DIS), which adopts the DCE service to the service interface [6] seen from the MAC layer and to the overall radio fieldbus system architecture [8]. The structure is 5m 4.5m 4m Rx A B C D E F G K 1st floor measurement 2nd floor measurement 3rd floor measurement X L M N W U 45,75 m Figure 2: Plan of the Magdeburg site J I T V P Q H O R S

3 B. Overview of the channel propagation characteristics The channel parameters, mentioned in Section A are measured with the use of a correlation type channel sounder based on DSSS technology. For each measurement configuration both static and dynamic measurement profiles are obtained [11]. In the dynamic measurement procedure, the transmitter is moving randomly in the close proximity of the selected point and averaged to produce a single impulse response profile. From the derived profiles, the average and RMS delay spread, the coherence bandwidth and the path loss are calculated. Table 2 presents the exact values of the measured channel parameters for static mode in several indicative points (A, I W). Channel Attenuation (db) Average Delay (ns) RMS Delay (ns) Coherence Bandwidth (MHz) Type A LOS-LC I LOS-LC V LOS-LC P LOS-HC Q LOS-HC J OP-LC M OP-HC N OP-HC W OP-HC Table 2: Channel parameters measured in static mode Concerning the attenuation, values between 60dB and 70dB represents the majority of measurements regardless the LOS or OP propagation environment. The attenuation seems to be more dependent to the distance between transmitter and receiver that the type of propagation. For the average and the RMS delay spread it is concluded by the measurements that are more dependent on the type of surrounding clutter than the existence of LOS. The majority of the RMS delay spread values range between 60ns and 90ns. High values correspond to positions near a wall or near the ceiling. Heavy obstruction propagation seems to have significant impact on coherence bandwidth results as well. The majority of the measurements are characterized by small values. Furthermore, measurements obtained at the third floor near the ceiling have even lower values. The recorded results indicate a difficult propagation environment and justify the choice of systems employing anti-multipath techniques. C. BER Measurement Procedure The modem used for the BER measurements employs DSSS transmission technique (REKA120 provided by ST2E). It has a data rate of 1.024Mbit/sec and it does not include any error-correction mechanism. According to its specifications, it can handle environments with RMS delay spread up to at least 100ns since it employs anti-multipath techniques (RAKE receiver). Though for the specific environment, the maximum measured RMS delay spread is 120ns, it is taken for granted that its RAKE receiver can handle satisfactorily these multipath reflections. In addition, as it uses external antenna, measurements at different received power levels are feasible by using external attenuators. For the BER measurements a BER tester is used (TE812 by Tekelec Temex) which can also be configured as data source. Apart from different received power levels, the use of directional and omnidirectional antennas, with the presence or not of adjacent channel interference is tested. The output power of the transmitter is adjusted to 18dBm. In order to control the received power level, a tuneable attenuator is used at the receiver. At the transmitter side TE812 is connected as data source whereas at the receiver another BER tester is used to measure the bit errors. In addition received power level is being recorded by measuring the Received Signal Strength Indication (RSSI) voltage. BER and RSSI measurements are collected for each point at the site. D. Static BER measurements with omni-directional antennas The receiver is placed at point Rx and the transmitter is positioned successively at the pre-selected points (Figure 2). In this configuration omni directional antennas with a gain equal to 2dBi are used and placed at a height of 2m at both the transmitter and the receiver. To estimate BER with a certain precision and a certain degree of confidence, one measurement has to be taken during a fixed time of approximately 100 seconds. If for the expected time of measurement no error occurs, it means that the BER is lower than the estimated BER. BER measurements are carried out for different receiving power levels. Four cases are examined: transmission at full power, transmission at half power, transmission at 10-15dBm above receiver s sensitivity (approximately 80dBm), transmission close to receiver s sensitivity (approximately 95dBm). B.E.R 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E Received power (dbm) Figure 3: BER vs. received power for points I,Q,N For the first three cases the BER performance is better than 10-6 (no error occurs during the I Q X

4 measurement time). Only for levels close to receiver s sensitivity, we have a noticeable BER worst than Figure 3 depicts three typical BER curves obtained from LOS-HC (Q), OP-LC (I), OP-HC (X) characterized points, close to receiver s sensitivity. E. Static BER measurements with directional antennas In this section the impact on reception of using directional antennas is examined. More specifically, we examine the BER performance of the modems using a directional antenna with a gain of 9dBi. Two cases are investigated: receiving power 10-15dB above sensitivity and receiving power close to receiver s sensitivity. The transmitter uses an omni-directional antenna and is now positioned at point Rx. On the other side the receiver uses the directional antenna and is positioned at the preselected points mapped in Figure 2. The direction of the receiver s antenna is decided by turning the antenna around a complete circle, choosing the direction that gives the highest RSSI measurement. Figure 4 depicts BER vs. received power for point I (OP-LC). Comparing the results of the two BER measurements cases, it is concluded that the usage of directional antennas can improve performance by reducing the required SNR for a specific BER. This is due to the reduction of multipath effects resulting from the directivity of the antenna. For example, for 10-6 BER the required SNR when directional antenna is used is improved by 8dB. B.E.R 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E Received power (dbm) omni directional Figure 4: BER vs. received power for point I for omni-directional and directional antenna F. BER measurements with adjacent-channel interference A jammer is placed at point H, the transmitter at point Rx and the receiver at several points in order to define borders of performance deterioration under adjacent-channel interference. The measurements are carried out first using omni directional antennas. The modem at point Rx is transmitting at a frequency of 2.464GHz, while the jammer is transmitting at 2.438GHz. As expected, the receiver functions satisfactorily over a wide area. Only when located very close to the jammer (distance less than1.5m) the receiver has degradation probably due to front-end saturation. Then, in order to increase the interference, the transmitting frequency of the jammer is changed to 2.459GHz, so as to partly overlap with the useful channel. The BER measured at points D, C, E, F, in this case are worse than However, when the same measurements are repeated at the second and third level they give BER values less than 10-6 at each point. This due to the fact of the radiation pattern in elevation which results in reduced received power from the interferer. Subsequently we replace the omnidirectional antenna of the receiver with a directional and we perform the same measurements as above, while the direction of the antenna is determined by the highest RSSI value. At the 1 st level the receiver is situated at points A, C, D, E, F, H respectively and the static BER measurements are obtained. The results show that the BER is lower than The same results are obtained for the 2 nd and 3 rd level as well. F. Further Measurements Apart from the stationary measurements, dynamic measurements are obtained as well with and without jamming interference. During these measurements the BER is lower than It is obvious that the system can overcome the fading of the channel during the movement. PER measurements are performed using two commercial modems. Since both of them implement the full stack of IEEE b standard, BER measurements are not feasible. PER measurements are obtained, using typical industrial packet sizes (6, 30 and 250 bytes). Both modems performed with a PER value below 0.001%, proving the effectiveness of the anti-multipath techniques implemented in these commercial products. V. DIS PERFORMANCE In order to point out the real-time performance of the proposed radio physical layer solution, a comparison with a wired solution will be presented. For this, the Profibus standard (EN Vol. 2) is selected. The basic equations relative to the Profibus performance are presented in (1) and (2), according to the EN50170 Standard [3], where: T MC : is one message cycle time (requestresponse). T S/R : is the transmission time of the request telegram. T SDR : is the station delay of the responder, defined as the time which may elapse between the receipt of a request frame s last bit until the transmission of a following (response) frame s first bit. T A/R : is the transmission time of the acknowledgment or response telegram. T ID : is the idle time of the initiator, defined as the time which expires at the initiator after receipt of a frame s last bit until a new frame s first bit is allowed to be transmitted.

5 0 1 9 T SR : is the system reaction time, which, roughly, defines the minimum period of changing values in a scanned slave. Message Cycle T MC =T S/R +T SDR +T A/R +T ID +2T TD (1) System Reaction T SR = Σ number_of_slaves T MC (2) For the cycle time calculations, a zero transmission delay (T TD ) and a zero message retransmission assumptions were made, while the standard 1.5 Mbps RS485 mode was selected. Moreover, two kinds of the message cycle were considered, one with a response of the maximum data length T A/R (246) and one with a response of the minimum data length T A/R (1), while the request is always with no data part. For the System Reaction time, the scenario is to have one master polling ten slaves, one slave with the maximum length response and 9 slaves with the minimum length response. The timing calculations, regarding to the standard RS485 Profibus case, have as follows: Tmc (µsec) RS485 Profibus 1.5Mb DSSS Profibus 2Mbps T MC (246) T MC (1) T SR Table 3: Profibus over DSSS PHY calculated times in msec The results are presented in Table 3, which shows the absolute timings (in msec) of the Profibus message cycles for the 1 byte (T MC (1) ) and the 246 byte (T MC (246) ) scanned variables for both the standard 1.5Mbps RS485 case and the 2Mbps DSSS case. Baudrate = 1.5 Mbps -> t Bit = µsec T S/R = 6 bytes * 11 bits = 66 t Bit T A/R (246) = ( ) bytes * 11 bits = 2805 t Bit T A/R (1) = (9 + 1) bytes * 11 bits = 110 t Bit T SDR = mint SDR = 11 t Bit T SDI = mint SDI = 37 t Bit Response Data Length (#bytes) RS Mbps DSSS 2Mbps Standard Message Cycle times (T MC ) T MC (246) = 2919t Bit =1.947 msec T MC (1) = 224 t Bit = msec Figure 6: Profibus T MC over 2 Mbps DSSS Standard System Reaction time (T SR ) T SR = T MC (246) + 9 * T MC (1) = msec The radio physical layer frame [12] is presented in Figure 5 and consists of a DCE related preamble part with a duration of 54 µsec, a DIS Header part of 72 bits and a MAC Layer PDU part with a limit of bits, in general, and 2040 bits in the Profibus MAC case, in particular. DCE Preamble DIS Header MPDU (User Data) 54 usec 72 Bits Bits Figure 5: Radio Fieldbus PhPDU structure In the case of the T MC calculations and for a 2 Mbps DSSS PhL, the overhead is = 90 µsec. Therefore, we must add this 90µsec overhead in each Profibus telegram transmission, but this duration along with the inter-frame-space time (IFS) overlaps with the T SDR and T ID timing requirements of the Profibus protocol, leading to a usage of the following values: IFS = 2 * Rx/Tx turnaround time = 10 µsec T S/R = 6 bytes * 8 bits = 48 t Bit + 90µsec T A/R (246) = 2040 t Bit + 90 µsec T A/R (1) = 80 t Bit + 90 µsec T SDR = max(11t Bit 100µsec, IFS) = IFS = 10 µsec T SDI = max(37t Bit 100 µsec, IFS) = IFS = 10 µsec Figure 7: Profibus SRD transaction through a DSSS link Moreover, in Figure 6, the T MC times are presented for both wired and DSSS Profibus and for the complete range of data telegram lengths, while Figure 7 depicts a snapshot from a Profibus DSSS-link implementation [13], and a 50 byte Sent and Request Data (SRD) Profibus transaction between a 1.5 Mbps wired (RS485) master station and a 2 Mbps wireless (DSSS) slave station. In Figure 7, the upper signal waveform corresponds to the wire medium, while the second one corresponds to the radio medium transmission / reception. The DSSS-link acts as a cut-through repeater between the air-medium and the wire, while, in the time between the request and the response frames (96 µsec) we have to add the 54 µsec of the DCE preamble, which is not shown in the figure, leading to a total of 150 µsec response time of the wireless slave node to the standard wired master.

6 VI. CONCLUSION In this paper, a physical layer for industrial radio fieldbus networks is proposed. It was demonstrated that DSSS technology employing RAKE receiver techniques can handle satisfactorily harsh industrial environments, while directional antennas can be used to compensate in cases where RMS delay spread increases dramatically. At the same time, the overall solution leads to a proper handling of the MAC layer PDUs, with transfer properties which result to comparable delay performance characteristics with existing wired solutions. VII. ACKNOWLEDGEMENTS We thank ST2E, Square du Chene Germain, Cesson Sevigne, Rennes, France, for providing the channel sounding equipment (SARACOM), the DSSS modems and the BER testers. VIII. REFERENCES [1] RFieldbus project IST , Deliverable D1.1, Requirements for the RFieldbus System, Technical Report, Apr [2] RFieldbus project IST , Deliverable D1.2, Assessment and Selection of the Radio Technology, Technical Report, Sep [3] Koulamas C, Lekkas A. Kalivas G., Koubias S., Papadopoulos G, Delay Performance of Radio Physical Layer Technologies as Candidates for Wireless Extensions to Industrial Networks, Proceedings of the 8 th International Conference on Emerging Technologies and Factory Automation, ETFA 2001, Antibes, France, Vol 1, pp [4] Juha Kalliokulju et.al, Radio Access Selection for Multistandard Terminals, IEEE Communications Magazine, October 2001, pp [5] IEC Ed.3:2000, IEC Standard , Fieldbus Standard for Use in Industrial Control Systems Part2 Physical Layer Specification and Service Definition. [6] General Purpose Field Communication System, Volume 2 Profibus, European Norm EN 50170, [7] RFieldbus project IST , Deliverable D2.1.1: Physical Layer Specification Part 1: Service Definition. [8] RFieldbus project IST , Deliverable D1.3: General System Architecture of the R- Fieldbus. [9] S. Kim, H. Bertoni, M. Stern, Pulse Propagation characteristics at 2.4 GHz Inside Buildings, IEEE Trans. Veh. Technol, VOL 45, No. 3, August [10] T. Rappaport, Indoor Radio Communications for Factories of the future, IEEE Commun. Mag., pp15-24, May1989 [11] R.J.C Bultitude, Measurements of wideband propagation characteristics for indoor radio with prediction for digital system performance, Proc. Wireless 90 Conf, Calgary, Alberta, Canada, July 1990 [12] RFieldbus project IST , Deliverable D2.1.1: Physical Layer Specification Part 2: Protocol & Operational Characteristics Definition. [13] RFieldbus project IST , Deliverable D2.1.2: Radio Interface Implementation.