Bluetooth Wireless Monitoring of Marine Propulsion Systems
H.A. Thompson and C. Rathi
Rolls-Royce University Technology Centre in Control and Systems Engineering
University of Sheffield
Mappin Street, Sheffield, S1 3JD
United Kingdom
Abstract
Over the past 10 years there have been rapid advances in wireless technologies. The availability of low cost mobile telephone technology and wireless interconnect for PC and electrical goods driven by the huge office IT and consumer electronics markets is creating interest in the adoption of this technology in other areas such as monitoring and control systems. Already products are appearing on the market which exploit this technology and many applications exist. This paper presents ongoing work at the Rolls-Royce Univeristy Technology Centre into potential exploitation of wireless technologies in aircraft, marine and industrial systems. In particular, this paper discusses marine application of Bluetooth for on-board systems, remote operational monitoring and control.
Keywords Wireless, Monitoring, Distributed Systems
1 Introduction
Wiring on an ship is complex, expensive, heavy and a key source of faults which lead to systems unavailability. From an installation point of view wiring requires considerable planning to allow routing throughout the ship passing from compartment to compartment through bulkheads that in turn need strengthening to meet structural demands. The wires themselves are a source of susceptability to failure as they can be damaged through accidents or in naval applications through battle damage. The marine environment is also harsh and the problems of open and short circuits, as well as intermittent faults from degraded contacts inside connectors, caused by vibration are well known. The aging of wiring is also an increasing maintenance ssue with more compelx systems and ship lifetimes of typically 30 years or more. Engineers thus expend great efforts to try and minimise wiring.
2 Overview of Bluetooth
The basic Bluetooth technology is an inexpensive ($3), low power, short-range radio on a chip that can be embedded in any device to translate digital data from computers. It is finding wide application in computers, PDAs, phones, cameras, etc. with enormous market predictions [1,2]. There are over 2000 companies working on applications. The radio sends and receives voice and data signals generated by other Bluetooth radios that come within the broadcast range (typically 10 metres). Because radio waves pass through solid substances Bluetooth devices can communicate through walls and other barriers that stop competing technologies such as infrared. (Bluetooth devices also do not require line of sight to communicate). The antenna for Bluetooth products can be as little as a short segment of conducting track on a printed circuit card [3].
Fig. 1 Bluetooth Radio
The radio operates on the globally- available industrial, scientific and medical (ISM) unlicensed radio band with frequencies between 2.402 and 2.408 GHz using frequency hopping with 1600 frequency changes per second. A single Bluetooth connection uses 79 different frequencies with a channel separation of 1 MHz. Three voice channels are supported as well as data transfer. Each data packet is transmitted at a different hopping frequency. (Frequency hopping is used to provide noise immunity as packets corrupted at one frequency can be re-transmitted at another. (Note although this is also claimed to give added security the hops are predictable and it is possible to buy equipment that can synchronise to this). Non-voice data can be transmitted either asynchronously at 721Kbits/s upstream and 57.6 Kbits/s downstream or synchronously at 432.6 Kbits/s in both directions. With control and protocol bits this is equivalent to 1 Mbits/s.
The connection itself is controlled by the Bluetooth link controller which is part of the baseband IC. This controls the protocol and link access routines. In the beginning all nodes in the piconet are in stand-by mode. Every 1.28 seconds they listen for a signal. If a signal is detected the Bluetooth module will look for a partner on 32 frequencies by sending telegrams. The maximum delay in contacting a slave is 2.56s although in most cases only half of this time should be needed. An inquiry telegram is then used to identify Bluetooth devices with the new unknown address. If no data is transmitted the master can tell the slaves to go into a hold mode. This is also the case when the master wants to talk to another piconet using the scatternet. Within 2 seconds individual Bluetooth controllers within a piconet identify themselves using a unique 48bit serial number. The first device identified becomes the master and controls the 1600 frequency hops per second. All users participating on the same piconet synchronise to this hopping sequence. Several piconets can communicate with one another over a “scatternet” with communication occurring between the piconet masters. Each of Bluetooth's picos, or cells, can accommodate up to eight PANs within a 100-square-meter range. Bluetooth works with short data packets and so the full duplex data rate within a scatternet that has 10 fully loaded piconets is 6Mbits/s.
3 Bluetooth for Ship Monitoring
In the past few years Bluetooth communications have become more robust and there is great interest in the use of Bluetooth wireless communications for system monitoring. Instrumenting a system is a time consuming task requiring considerable wiring. Considering a high speed merchant vessel there is interest in monitoring the full propulsion chain consisting of the gas turbine engines, the gearbox and shaft and the steerable waterjets. This requires a range of sensors at different locations in the ship operting in diverse environments. The ability to simply wire a sensor to a Bluetooth transmitter and then collect the signals via radio would be a great advantage. Of course a concern is the ability of the signals to propagate throughout the ship which is predominantly made of steel, however, recent work [4] has shown that Bluetooth propagates well within the confines of shipcompartments and indeed it can propogate through hatch seals and insulation around pipes allowing transmission through non-conductive bulkheads from compartament to compartment.
Fig.2 High Speed Merchant Vessel
With suitable location of only a relatively small number of Bluetooth transmitters it would be possible to provide a wireless communications network that would cover the entire ship. Due to the distances involved this would require exploitation of both the piconet and scatternet functionality of a Bluetooth network. Recorded data could be downloaded via a PC or PDA as the crew walked around the ship a via base stations connected to an intranet. This would significantly simplify both the installation and maintenance of systems.
Fig. 3 Typical Gas Turbine Engine Wiring Harness
At the local level it would also be able to simplify the wiring on individual plant. Fig. 3 shows the complexity of the wiring on a modern marine gas turbine engine. Significant reductions in complexity could be possible resulting in higher levels of availability.
4 Data Rates and Environmental Considerations
Although a promising technology with lots of potential there are still a number of limitations which need to be considered before Bluetooth will become accepted. Considering the typical parameters that need to be monitored for a packaged gas turbine engine, with gearbox, shaft and waterjets, the sample rates required vary from high bandwidth (above 20KHz for vibration monitoring), medium bandwidth (around 10KHz for pressures and speed), low bandwidth (less than 1KHz for temperatures, oil monitoring, fuel flow, fuel, oil and hydraulic pressure, throttle position, and oil quantity) and discretes (valve position open/shut).
Fig. 4 Packaged Gas Turbine Engine
Although the bandwidth of Bluetooth is nominally 1Mbit/s it is important to consider what sample rates are achievable. Bluetooth is a standard developed for consumer electronics. It is designed to operate with voice and data packets. There are 6 types of data packets that can be specified DM1, DH1, DM3, DH3, DM5 and DH5 transferring increasing amounts of data from 17 bytes per message to 339 bytes per message.
(The DH versions operate slightly faster due to the fact that some checking is switched off.) Considering a system composed of individual sensors producing 12bit data each with their own RF interface transmitting DM1 packets at most a packet can be transmitted once every 1350 microseconds and so around 800 packets can be sent a second.
Typically then commercial Bluetooth systems, can thus only sample in the millisecond range confining use to low bandwidth and discrete position monitoring if no pre-processing is used. Indeed early work within the Rolls-Royce UTC concentrated on the use of Bluetooth for relatively slow signals using modules supplied by Crossbow Technologies [4].
Consideration of the protocol overheads if each sensor had it’s own Bluetooth interface and sends it’s own packet is even more illuminating. For every 12 bits of data sent there would be an overhead of
ACL data packet = 72bit access code, 54 bit header, 16bit CRC + 17bytes data (for DM1)
giving 278 bits. This equates to 96% of the bandwidth being lost due to the communication protocol. This immediately highlights the fact that in a practical system sensors would need to be grouped together with a Bluetooth interface. Even with 8 sensors wired into a single Bluetooth transmitter the protocol overhead is still 65%. To get the best out of the system careful optimisation of data flow is necessary.
It is also important to consider the environment in which a Bluetooth systems must work. Depending on location the modules may be subjected to high temperatures with exposure to vibration, flexing, stress, solvents, lubricants and fuel. On the core of a gas turbine engine temperatures may be above 5000C requiring high temperature wire and connectors. Although there is an interest in using wireless systems in these areas it is difficult to find electronics that will survive in the environment. Thus initial applications of wireless systems are confined to relatively benign environments.
5 Wireless Vibration and Temperature Monitoring
As highlighted to monitor higher bandwidth signals it is necessary to perform pre-processing of the data. For this the University Technology Centre has more recently been investigating the use of a system produced by OCEANA technologies for vibration and temperature monitoring. This uses an Analog Devices ADSP-21065 32 bit floating point processor (66MIPS) for digital signal processing and data storage. This does Fourier analysis and feature extraction prior to wireless transmission. An ARM processor is used for dealing with data transmission.
Although there are concerns about the environment in which the system must survive in the longer term of particular interest in the first instance was the performance of the sysem in an electrically noisy environment. If the system cannot cope with this then other environmental considerations are not an issue. To provide a difficult environment an electrical machine was chosen as the test vehicle and the system was installed on an integrated starter generator rig. Two vibration measurements were taken (horizontal and vertical) and 4 temperatures were monitored. These were the motor temperature, the temperature of the water cooled casing, the enclosure temperature and the ambient temeprature. This system is shown in Fig.6.
Fig. 6 Integrated Starter Generator
The aim was to use the system to monitor vibration and temperature characteristics as the system was run in test and look at the reliabilty of the system considering issues such as interference and fading. The wireless transmitter was located with the power electronics of the system to provide an extreme electrical environment in which to operate. The Bluetooth signals were captured on a portable PC located outside of the test cell.
Fig. 7 Bluetooth Transmitter and Power Electronics
6 Results
The first and most impressive result was the reliability of the communications in the noisy environment. The system was found to be extremely robust with no problems of interfaerence or fading. Some results from the system are shown in Figures 8 and 9. Fig. 8 shows the temperature profiles of the system as it is run up. From the traces nosie can be seen on the signals, however, this is not significant and can be removed via filtering.
Fig. 8 TemperatureMonitoring
A vibration trace for the horizontally mounted accelerometer is shown in Figure 9 with the system running at 1600 rpm.
Fig.9 Vibration Monitoring
7 Providing Power
There is no real advantage to removing signal wiring unless the need for power wiring is also removed. When transmitting a Bluetooth device should consume 8-30 milliamps (or less than 1/10th of a watt) and in "hold" mode around 30 micro amps. To conserve power Bluetooth also has sniff and park modes. In sniff mode they can scan at a lower rate dictated by the application. In park mode (where the least power is used) the slave module is synchronised within the network but does not transmit unless a wake up call is sent. Typically, Bluetooth modules only require about 2mW in the power saving modes making them suitable for battery operation. Standalone sensors could use batteries that would last for approximately 2-3 years. Battery technology is likely to improve but for future fit and forget, low maintenance systems alternative sources of power are required. Already work has been performed on energy harvesting to convert heat and vibration energy into power for the sensors (studies at MIT have shown that around 20mW is available on a gas turbine engine).
8 Bluetooth in Control Systems
Bluetooth for system monitoring is a viable concept [6.,7], however, could it ever be used in feedbackcontrol systems. Already, companies such as ImpulseSoft, Con1, connectBlue and CSR all provide Bluetooth equipment for industrial applications. The products, however, are targeted towards providing Bluetooth access points onto other networks such as Ethernet, Controlnet or Profibus. Some work has been done by ABB in Norway 15 investigating Bluetooth performance in a chemical pulp factory [8] but for safety-critical control certification becomes a big issue and the consequences of interference that results in the control being interrupted or lost could be catastrophic. Microwave ovens have been shown to interfere with Bluetooth if very close and other typical activities on a ship such as welding may well cause problems.