Airfield Ground Lighting Automation System Realised With Power-Line ommunication * GmbH Unterschlauersbacher-Hauptstr. 10, D-90613 Großhabersdorf, Germany Phone: +49 9105 9960-51, Fax: +49 9105 9960-19, Email: Gerd.Bumiller@iad-de.com Gerd Bumiller *, Nils Pirschel ** ** Siemens AG, I&S ITS A Schuhstr. 60, D-91052 Erlangen, Germany Phone: +49 9131 7-21033 Email: Nils.Pirschel@siemens.com Paper ID: KE 1132 Affiliation: T17) Experimental Systems and Field Trials Keywords: Field Trials, OFDM, Airport Solution, Automation Abstract The Airfield Ground Lighting Automation System (AGLAS ) is suitable for individual switching and monitoring of airfield lighting in order to meet latest IAO (International ivil Aviation Organisation) and FAA (Federal Aviation Authority) recommendations, to enhance safety of aircraft ground movement in particular areas of aerodrome (stopbars, lead in, lead on) and to built up the basis for an A-SMGS (Advanced Surface Movement Guidance and ontrol System). Many existing airports world-wide have the need to be updated with individual switching devices. Power-Line based systems which must be able to communicate reliably on different kinds of cables and isolating transformers power supplies using different kinds of thyristor controlled regulators are limiting at least infrastructure costs for the airports. AGLAS has been developed since conventional power line based communication systems have problems to prove outstanding reliability and availability under all circumstances as well as do not keep pace increasing requirements regarding overall performance. 1. Introduction At the last ISPL 2002 in Athens the PL community voiced the wish to learn more about realisations of transmission systems and the respective field trials. After realised a Power-Line system on an extraordinary power network together with Siemens AG last year and has the results of the field trials available, we would like to consider the wish of the PL community and present these here. 2. Representation of the System Transmission in AGLAS originates from one PL master and goes on to the remotes over the serial current circuit and a number of transformers Tower onstant urrent Regulator LAN PL 0.4-6 kv 2.8-6.6 A 3 15 km length, 10 to 140 remotes per ring figure 1: structure of AGLAS At nearly every airport the constant current regulators, cables, transformers and lamps already exist. The newly installed system has to accept the rough environment of the thyristor controlled regulators, the length of the ring between 3 to 15 km with shielded and non shielded cables, partly grounded at a lot of places and transformers built for a peak current of 15 amps, but not designed for communication. The number of remotes and transformers inside of a ring varies between 10 and 140 for some special application at the US market up to 220. Since this system is needed at airports especially in weather conditions according to AT III (e.g. fog), a switching command has to be executed within 1 second and acknowledged at the tower within another second. While the switching command can easily be sent to all remotes via broadcast within the given time, a response from all remotes involved is necessary and within the
required time critical. The usual polling of the participants is not sufficient here to meet the requirements. Each malfunction of a lamp is diagnosed and reported immediately. The behaviour of a lamp in different failure situations needs to be individually configurable and remains changeable. The remotes are installed underground at the airfield and have to meet the requirement IP 68 (pressurised water proof up to 2m). As ventilation is not possible for the electronics the power dissipation of the remotes has to be minimal. This limits the possible transmission power and asks high efficiency of the power supply. While the PL master is installed in the primary circuit and requires a voltage sustaining capability of > 6 kv, the remotes are installed in the secondary circuits between the transformers and lamps. The PL master communicates with the tower using a data network. A constant-current regulator that delivers a current of 2.8-6.6A requested by the tower supplies the circuit. The pilots already perceive a variance in the current flow of 10 ma as a different intensity of light. The different kinds of thyristor controlled regulators produce a lot of interferences in the power network. A filtering unit has been developed for the PL master, that blocks the interferences and if necessary can be combined with further filter blocks. These filter units have to be usable for peak currents of 15 A and must not cause any nonlinear distortions for communication. ommunication between master and remotes or between remotes always runs over transformers. The transformers used at airports are soft magnetic, which prevents the failure of an entire circuit when only one lamp is defective. The transfer ratio is nearly 1:1, which is why in the primary circuit the current is about the same as in the secondary circuit. The transformer and the current circuit are not designed for data transmission. Therefore, the wave impedance of transformer and power line differ dramatically and high reflections and attenuation are caused. For instance some transformers at 125 khz [1] have impedances > 500 Ω and a insertion loss of up to 18 db. A direct communication of the master with all remotes in long circuits is not possible. In order to minimise the number of necessary repetitions of a message on the circuit, the system has to be designed for a high dynamic range in the signal amplitude. The maximum transmission power is limited by the dissipation of the remote. Thus the high dynamics can only be achieved in an environment with very low interference. While the regulator s interferences are filtered in the master and halogen lamps cause low interference, the power supplies in the remotes are a strong source for interference already. Further interferences are caused by crosstalk between cables, injected interferences and other power supplies when the same circuit supplies further electric devices. 3. Transmission Technology This system developed by the company and Siemens AG is based on the ASI presented on the ISPL 99 as Integrated PL-modem based on OFDM [2] with powerful forward error correction and possible data rates up to 100 kbps. figure 2: DL-2A ASI The DL-2A ASI of is a mixed mode design. The complete analog signal path, except the power amplifier, in send direction is integrated in this ASI. The intensive usage of digital signal processing and forward error correction results in an optimally adapted system, which cannot be realised in this form with the customary available components. The intensive discussion of the concept for transmission is done in this year s ISPL paper System Architecture for Power-Line ommunication and onsequences for Modulation and Multiple Access by Gerd Bumiller [3]. A compromise had to be found for the transfer parameters regarding the following points: The frequency bands to be used ranges between 10 and 150 khz The transmission channel has a strong low-pass character and the lower frequencies are taken into consideration. The transfer blocks should be as short as possible because of the real time requirements. The applied network management requires at least 32 bytes in a symbol.
For a secure transmission the redundancy for the FE should be very high which results in a low bandwidth efficiency. The remote as sender is limited by the transmission power not by the spectral power density and therefore the send energy of a symbol increases with the length of a symbol. As many independent channels as possible. Pulse noises are the main interference. Because these requirements partially contradict each other the parameters have been chosen after weighing the individual arguments and evaluating the first field trials. In AGLAS the data consumption is limited. But for new functions of the Advanced Surface Movement Guidance and ontrol System higher data rates are necessary. It is useful to select slower data rates as the maximum of the DL-2A chip with 100 kbps. The reliability of communication grows, fewer repetitions are necessary and more independent frequency channels are available. The requirements concerning crosstalk will be considered separately. 4. rosstalk At an airport a great number of these serial current circuits are parallel. Two or more circuits on which the lamps alternate, and that are supplied by separate regulators in sometime separate transformer stations for safety reasons, for example, supply lamps like the Runway-entre-Line. (figure 3). R GPS PL PL figure 3: PL system on the runway centre line R GPS Additionally the cables of many signal lamps are often lead parallel through the same cable ducts over great distances. The cables used for serial circuits are sometimes shielded, but sometimes they are not. Depending on the applied grounding concepts sometimes a lot of crosstalk may occur. A separation of the communication by frequency alone is insufficient due to the characteristics of the channel that sometimes make only few channels available for transmission and also the high dynamics of the signal amplitude that the system was designed for. To alleviate this problem, the possibility to additionally separate the systems in the time domain via timeslots has been introduced. As the real-time requirements to the system are very high, the single timeslots have to be as short as possible. They have been set to the duration of a transfer symbol. The remotes in a system synchronise to their master or the remote that serves as a repeater. In order to keep the timeslots of each frequency band with several masters the masters have to be synchronised to below 100 µs. Over short distances this can happen on a synchronisation line or over greater distances via GPS (Global Position System). To test the functionality a serial current circuit with 80 remotes has been equipped with 2 masters directly following each other. One master s filtering unit was removed so that both masters were sending and receiving on the same circuit. The remotes have been assigned half and half to the masters, and the synchronous communication of both masters with their remotes was examined. Since the identical medium was used, this test case equals a worst case of crosstalk with 0dB attenuation. In this case it was proven that time slots are kept reliably and communication is ensured. 5. Network Management The network management applied here is based on that presented on the ISPL 2000 as A Network Management System for Power-Line ommunications and its Verification by Simulation [4]. It had to be altered in several areas to meet the system requirements. ommunication works on a master - slave basis where every client in the network is a possible repeater for the other participants. Since at some airports the location of a remote within a circuit is unknown automatic routing is needed. If a remote that worked as a repeater for another fails, an alternative route will be found automatically. The remotes analyse the entire data traffic and form tables. In these tables are the remotes, which could work as a repeater for the remote. The best four in this table are transmitted by the remotes to the master, together with a weighting, during cyclic polling. The master combines this information with its own tables that contain information on how well it reached the respective remote on the different routes. The master calculates the best route for every packet to be sent and also for every repetition. Since the effort increases exponentially with every repeater level and the data in a
transfer block are strictly limited, the system was designed for a maximum of two repeater levels. Upon the installation of a remote it is not yet assigned to a master and does not know the frequency band used by the master. In that phase a remote scans all frequency bands until it is directly addressed by a master and logs in with that master. From then on it will use this frequency band. However the remote can be requested to return to the original state of frequency scan. When that occurs all routing tables are erased. Since the master does not know the sequence of participants in the circuit log in works by a repeatedly performed trial-and-error process. With every iteration more participants become available that can serve as repeaters for new participants. The routing information at that stage is not yet very precise because the routing tables are just being built. After logging in every remote is provided with its configuration data, that will determine the behaviour in various situations. After all participants are logged in and the single configurations are loaded the circuit can change into operation mode. This entire process is automated in the PL- and only needs to be initiated after the installation of all the components. ompared to systems where the routing has to be determined manually, this is a very short start up time. The challenge to the network management is the fast switching of the lamps and the according response from every remote whether the procedure was successful or not. For the reliable broadcasting of the switching command the master determines the number of repeaters and repetitions necessary according to the current routing information. The worst case is, if every remote is infected by a switching command. Since the real time requirements are very strict, the response system cannot work based on data blocks. The plain information content of a response is only one bit. If shorter timeslots and the position of every lamp have been assigned for this case in advance, this information can be transmitted by the sending / not sending of a preamble. Then the duration of a timeslot can then be reduced dramatically. The method provides enough redundancy because the synchronisation works already at SNR of 0 db. The remotes that acted as repeaters in a broadcast now collect the responses and transmit them as a data block to the master. The master combines the information and can report the result to the level above and also directly address lamps that have not been switched and force-switches them. As an update of the software for such a complex software is to be expected, but exchanging all the remotes at an airfield would be extremely expensive or impossible, the possibility of a software download over the power line has been stressed especially. That was designed in a way that even the transmission parameters can be changed and the software update takes place in all remotes simultaneously. In fact after the first field trials at the airport Leipzig / Halle with more than 100 remotes the transmission parameters including synchronisation and bandwidth of the communication method had to be changed, which can only be realised by a software update. For all remotes, except one that had to be updated later on, the entire successful software update took no longer than 1.5 hours. 6. Procedure and Problems in the Project The development of AGLAS system has been started in December 2001. All hardware components and great parts of the software had to be designed totally new. In July 2002 for the first time a lamp was switched over the complete communication route in a test set-up. In August 2002 the test set-up was installed at the airport Leipzig / Halle. Especially at that stage the elaborate measuring functionality, which is integrated in the DL-2A ASI and can be evaluated with special software, proved to be very helpful. No additional measuring devices were needed and also could not be used in that environment. The exact analysis of interferences and transfer functions helps to recognise and solve problems quickly. The last remaining weakness were interferences caused by the remote s power supply that could now be reduced drastically by a new power supply concept in a actual remote. Also the possible transmission range could be increased significantly. The development of this power supply was certainly problematic. For instance, after completion of the design, the remote with excellent transmission results had to be withdrawn again as it caused problems with the constant-current regulators and made the current vary up to 1 A. Up to now the new remotes have been installed at the airport Leipzig / Halle but the following results are based on the first design approach. 7. Results at the Airport Leipzig/Halle For the field trials at the airport Leipzig two current circuits were available. On the one hand that was the Runway Edge circuit (REH) with 42 remotes and a total length of 7100 m. After the incoming supply line of the transformer station there is one remote on every 120 m in the circuit. On the other hand we worked on the touch down zone (TDZ) circuit with 60 Remotes and a total length of 5800 m, whereas in this circuit there are 5 remotes in a shaft and the distance between the shafts is 90 m.
In both current circuits the transmission quality for the different frequencies were determined in Summer 2002. In both circuits only one repeater level was needed at frequencies up to 70 khz. Up to 110 khz all remotes could still be reached reliably with the two repeater levels. Above this frequency some remotes in the TDZ circuit could not be reached anymore. Reliable operation was guaranteed in several frequency bands. Via the available testing tools a complete routing statistics can be traced that determines how frequently certain routes are used, and besides the detailed listing also provides all necessary parameters for evaluation. In a detailed evaluation of this routing statistic it was shown, that the network management in fact always finds the best transmit path. For a more difficult test the REH-circuit and the TDZcircuit were wired in series at the transformer station. There a current circuit of 12900m length with 102 remotes was established. The location of the remotes in the circuit was rather unfortunate, because there was a long stretch without any remote in the middle part. ommunication in this circuit was only possible on frequency bands below 70 khz. 25% of the remotes were reached directly by the master, 35% over a repeater and 40% over two repeaters. However, depending on the situation the packet error rate for a few remotes ranged at 20 to 50% already. Therefore, reliable operation was not guaranteed anymore. An actual remote with less self-interference present much better results, which ensure that the requirements for a save and reliable communication under all circumstances of a circuit with 140 remotes stretched over 15km will be met. Also the response time for switching entire groups of lamps were measured in this large serial current circuit. The time between the switching command and the response to the tower varies over the situation but is clearly below the requirements. 8. onclusion The project described here shows that power-line communication is advanced enough to realise complex systems within a reasonable time and the technology can also be applied successfully in rough environments with high real-time requirements and safety standards. The actual transmission technology is important, in a project like that however actually only a lesser topic. All topics from communication structures at the airport to embedded Linux, and waterproof casing technology, and development of power supplies have to be mastered as they influence each other and can only be solved successfully in a well-adjusted team. Therefore the authors would like to thank all those who worked in the AGLAS project and helped make it successful. 9. References [1] F. Krug, Bericht zur Messung der Transformatoren, Lehrstuhl für Hochfrequenztechnik, Universität Erlangen-Nürnberg, 29. August 2001. [2] M. Deinzer, M. Stöger, Integrated PL-modem based on OFDM, GmbH, Großhabersdorf, technical paper for ISPL, 1999. [3] G. Bumiller, System Architecture for Power-Line ommunication and onsequences for Modulation and Multiple Acces, GmbH, Großhabersdorf, technical paper for ISPL, 2003 [4] M. Sebeck, G. Bumiller, A Network Management System for Power-Line ommunications and its Verification by Simulation, GmbH, Großhabersdorf, technical paper for ISPL, 2000.