Electronic control units for automotive electrical power systems: communication and networks

Transcription

1 1217 Electronic control units for automotive electrical power systems: communication and networks A Shrinath and A Emadi* Illinois Institute of Technology, Chicago, Illinois, USA Abstract: The last decade has seen a resurgence of advanced technologies being implemented in automobiles. Functions that were considered highly complex and difficult to be implemented are not only being provided but various other facilities are also being built on those very functions. Everything from engine control to multimedia and infotainment services are implemented in today s automobiles. This has been due largely to the strides made by the electronics as well as the communications industry. The emergence of the concept of providing the customer with all possible services has led to an explosion of implementation of high-end electronics, each providing a facility of its own. When so many devices are being embedded in a car, there also has to be a mechanism that should regulate the way in which data transfer takes place. It is the objective of this paper to provide an insight into most of the communication protocols that are being used, as well as technologies that are under development in the automobile industry of today. Various protocols and their modus operandi are explained, with schematics completing the picture. Keywords: automotive electrical systems, Bluetooth, communication, controller area network, electronic control units, network 1 INTRODUCTION The modern-day automobile was not built in a day. It took years and thousands of patents to build it. This has been an industry that has seen rapid growth in terms of its technical as well as its aesthetic aspects. The carbuilding effort has also been receptive to growth in other industries recently, most prominent among them the electronics industry. Until a decade ago, the automobile Fig. 1 Growth in automobile wiring industry suffered from a conservative approach towards computer-controlled processes. However, rapid strides made by the electronics industry made it possible to manutogether With the necessity of wiring the electronic devices facture smarter devices. By incorporating these devices, comes the issue of data management. This task car manufacturers were able to offer more services and is accomplished by a set of rules known as protocols. features in a car. But, this has come at a price. The cost These protocols govern the way in which data is com- of implementing such electronics amounts to almost a municated from one electronic device to the other. quarter of the total cost of manufacturing an automobile Examples of such protocols are x-by-wire, MOST, CAN, [1]. While electronic systems averaged $100 in an autohere TTCAN, SAE J1850, and Bluetooth. A noted exception mobile in 1977, they have averaged around $1800 until is Bluetooth, which is a wireless protocol. Details quite recently [1]. Coupled with the implementation of about this protocol are explained later in this paper. computing solutions for solving automotive problems The advantage of using these protocols is that the task is the fact that these electronic components need to be of communication between the devices is centralized (i.e. connected by wires. Figure 1 represents graphically the the task is handled by a central control unit). Thus, all growth in wiring in recent years [2]. the devices are networked together. The implementation of these networks not only solves the problems of The MS was received on 26 December 2003 and was accepted after traditional wiring systems, it also provides scalability; it revision for publication on 22 June * Corresponding author: Illinois Institute of Technology, 3301 South Dearborn Street, Chicago, IL , USA. emadi@iit.edu leaves open opportunities for adding new components in the future without any major changes to the network.

2 1218 A SHRINATH AND A EMADI It also makes it possible to integrate different networks. (SAE) Congress. The CAN protocol is based on the This will increase the functionality of already existent principle of broadcast transmission technique. It is systems. The high data rates available today (they vary basically a communication technique in which data to from a few kbps to a few Mbps depending on the netincluding be transmitted on a network is sent to all the nodes work used) make it possible to implement such seemnodes the one to which the data is intended. The ingly simple networks for real-time control systems. To look inside the data packet to see if the message implement these networks in the safety critical systems was meant for them. If not they simply discard the of a car will take some time, because a failure of these packet. The node to which the data was intended then networks might result in loss of human life, a matter of downloads the data and processes it [4]. very grave concern. Such networks are, however, already Some of the features that have made CAN so popular in use in many places in a car, such as multimedia, doors, are speed, data length, and event-triggered mechanism. seat adjustments, and trunk release. The speed of data transmission in a CAN network can According to current estimates, some 20 per cent go up to 1 Mbps. This is very helpful in real-time control of the total manufacturing costs for automobiles is systems that can afford very low latency. Thus, due to attributed to electronics. These electronic devices or the high speed offered by CAN, low latency and, hence, electronic control units (ECU ) are generally in a single time-efficient control can be achieved. CAN frames are chip, 8-bit microcontroller with around 100 B of RAM, also short in length, owing to which there is minimal 32 kb of ROM, few I/O pins to connect to sensors and delay in the reception of messages. actuators, and a network interface [3]. They can be CAN is basically an event-triggered mechanism. This broadly classified into two categories: body electronics means that the transmission of data is prompted only and system electronics [3]. when a specific event occurs. For example, data trans- Body electronic devices are responsible for those mission might take place when a button is pressed or a operations not directly related to the movement of lever pulled. Owing to this property, the bandwidth the vehicle. Examples include theft avoidance systems, available is made maximum use of because of the mini- air-conditioning control, audio equipment control, seat mum load put on the bus system. The combination of adjustment, window control, and air-bag control. Another high speed and short message length results, therefore, interesting concept is keyless access to the vehicle. The in a low delay, owing to which CAN can be implemented user simply has a card-like device. Sensors in the car in control systems that have a very low tolerance for detect the card and automatically unlock the doors. The delays. The addressing system is based on message user then sits and simply has to press a button to start identifiers rather than physical addresses for nodes. the engine, rather than the conventional way of putting Every CAN message has a unique identifier that assigns the key into the ignition and turning it. System electronic a priority to the respective message based on the binary devices, on the other hand, are directly involved with value of the identifier. A lower binary value is assigned the movement of the car. The most important impacts a higher priority and vice versa. Any node wishing to have been in computerized engine control and anti-lock transmit sends the message to the CAN controller, which braking systems (ABS). Another application that has in turn broadcasts the message. All the other nodes then seen the implementation of computerized control is get ready to receive and the nodes that have no use for transmission control. the message discard the message. The main advantage of using ECUs and networking them is that they give scalability and functionality. Automotive electronic devices not only replace mech- 2.1 CAN arbitration anical systems, they also help in integrating other systems. We shall now proceed to discuss the various protocols A problem that arises in a network is the problem of that are used for the control of the transmission of data collision avoidance. There could be a possibility that between devices. with so many messages going around in the network messages could collide, which could lead to a loss of messages. CAN overcomes this by a simple method known as carrier sense multiple access/collision detection (CSMA/CD). In this method, the nodes first listen to the 2 CONTROLLER AREA NETWORK (CAN) bus to see if it is free and then they transmit. It was in 1983 that Robert Bosch GmbH in Germany started an internal project to develop an in-vehicle network that would serve to connect all the electronic components in a car. In 1986, they introduced the concept of the serial bus-based controller area network (CAN) for the first time at the Society of Automotive Engineers 2.2 CAN error detection The errors that occur in a network may be at either the bit level or at the frame level. Fortunately, CAN is equipped to handle both kinds of error [5]. The major

3 ELECTRONIC CONTROL UNITS FOR AUTOMOTIVE ELECTRICAL POWER SYSTEMS 1219 Details cannot be released as these are proprietary technologies. Figure 2 graphically depicts how CAN is implemented in automobiles. types of error that can be handled by the CAN protocol are bit, stuffing, cyclic redundancy check (CRC), form, and acknowledgement errors. As mentioned before, CRC is employed to detect any errors occurring during transmission. Here both the receiver and transmitter would have stored a mutually agreed polynomial P(x). The transmitter then represents the data to be transmitted by a polynomial G(x). G(x) is then divided by P(x). The remainder is then transmitted in the frame to the receiver. The receiver then performs the same division to see if it gets the same remainder. If not, then Fig. 2 CAN system overview the receiver can infer that an error has occurred. 3 TIME TRIGGERED CONTROLLER AREA NETWORK ( TTCAN) 2.3 CAN architecture Communication protocols can be divided into two A basic CAN protocol consists of three layers: physical classes: event triggered and time triggered. Both these layer, data-link layer, and application layer. The data-link methods vary as far as their operating principles are layer (DLL) is implemented in an electronic component concerned [8]. Time-triggered-controller area network known as the CAN-controller. Several semiconductor ( TTCAN) is a communication protocol based on CAN, manufacturers, such as Intel, Motorola, and Philips, though the difference is that it uses a time-triggered make it. The CAN controller consists of the following: mechanism rather than the event-triggered mechanism CPU interface logic (CIL), which handles the data used in CAN. The specifications of TTCAN are defined transfer on the bus; bit stream processor (BSP), which in ISO standard It utilizes all the error detection handles the streaming of data between buffer and bus mechanisms as well as the soundness of CAN. line; error management logic ( EML), which is involved The event-triggered model is known as an asynchronous with error management; bit timing logic (BTL), which model whereas the time-triggered model is is responsible for the synchronization of bit streams; known as a synchronous model. In the event-triggered transceiver control logic ( TCL), which handles error model, data transmission takes place when a certain detection and correction, transmission, and reception of event takes place, for example when a button is pressed the data and arbitration; and message buffer memory or a lever pulled. Here data transmission takes place at (MBM), which is used to store messages for future random on the time line. However, in the case of a timetriggered transmission. Recently, a variation of the CAN protocol, model, data transmission takes place at specific called the CANopen, has been developed [6]. intervals on the timeline. In this mode, all the nodes are There are many ways in which CAN can be synchronized to a master clock so that all of them have implemented in automobiles. For our case, we shall con- the same sense of time. Each node is allotted a slot time. sider the various ways in which the Jaguar car company It is only during its allotted slot time that the node can has implemented CAN in its XK8 sports car [7]. The transmit. This is something akin to the time division applications in this car are the engine control module multiple access scheme. As mentioned before, trans- ( ECM), antilock braking system (ABS), transmission mission of data takes place due to the progression of control module ( TCM), instruments and driver information time. But, how do the nodes know when to start trans- module (INST), J-gate illumination module mitting their data? In the static scheme, nodes are (JGM), and intersuspension data transfer system. The allotted slots in time, but in the time-triggered CAN this following figure demonstrates amply the concept of problem is solved by the transmission of a special frame CAN implementation in automobiles. Data access points known as the reference frame. This is distinguished from are specially ear-marked from where data has to be other frames by virtue of its identifier. This reference collected. Appropriate sensors are wired to these data message has a bit called the start-of-frame (SOF) bit. access points, which are in turn wired to microprocessors The reception of this bit denotes the instance of time at that perform adequate data processing. The data is then which the data transmission can take place. The TTCAN placed on the CAN bus along with the medium access has two levels of operation defined in the specification. control mechanism. The data is then sent to the CAN Level 1 is by virtue of the property of the reference controller for control mechanisms to be implemented. message. This reference message ensures the time-triggered In a more generalized manner, CAN can be used to network operation of the CAN. Level 2 ensures the global time body control modules and infotainment networks. synchronization of the nodes in the network. The difference is that for level 1 implementation, the reference message carries one byte of information necessary for control purposes, whereas for level 2 implementation,

4 1220 A SHRINATH AND A EMADI the reference message contains about four bytes of con- length. A counter Cycle_Count is used to indicate the trol information with the other four bytes being usable number of the current basic cycle. It is incremented by for data transfer. one every time a basic cycle is completed. Apart from The advantage that it enjoys over CAN is that during the exclusive and arbitrating window, another window, its operation cycle, TTCAN permits the transmission known as the Tx_Enable, is used. This is contained of the regular time-triggered messages as well as eventtriggered within the exclusive and arbitrating windows. It is used messages. Certain time slots are reserved for to indicate to the node the time at which it is supposed the event-triggered messages. For example, x-by-wire to start transmitting. This is done so that the message systems, which integrate sensors and actuators for per- following the current one may not be delayed due to the forming certain critical real-time control functions, need delayed transmission of the current message. TTCAN. This leads to a predictable behaviour of the The time unit used in any network transaction in the network, which is very important owing to the distributed TTCAN network is known as the network time unit architecture of many networks. A distributed (NTU). It can be either a CAN-bit duration in the level 1 network consists of many subsystems organized into TTCAN or a fraction of a second when implemented in subnetworks. Each subnetwork may be implementing a the level 2 TTCAN. A message state counter (MSC) is different protocol. Therefore, a protocol that clearly attached to every message transmitted or received. An defines the transmission procedure to be followed at error to this message is responded to by an increment in well-defined time slots would clearly reduce the complexity the MSC. A node then flags an error if either the MSC of the network, as well as utilize the available reaches a value of seven or the difference between two bandwidth fully. In addition, it would reduce the time MSC is greater than two. required, however small, for the arbitration procedure Implementation of the TTCAN protocol is handled in the CAN that grants the nodes access to the bus by the TTCAN IP module, which implements not only for transmission of messages. It would also reduce the TTCAN protocol (ISO ) but also the CAN the probability of a message with low priority being protocol (ISO ) [10]. Bit rates of up to 1 Mbps restrained from transmitting that message for a long time. are achievable by this module. All the requirements of Many of the subnetworks in up and coming cars will a time-triggered network, including global time synbe highly crucial for the safety of the car, for example, chronization and clock-drift compensation, are provided the x-by-wire systems being implemented in cars. In these by this module. Message objects and message masks are systems, the control of the car would be decoupled from stored in the message RAM for operation in a CAN the process to which it is attached [9]. Therefore, in these network. For TTCAN networks the trigger object is cases, the deterministic nature of the network, as well as stored in the trigger RAM. The message handler handles its reliability to deliver messages to sensors and actuators all message-handling instructions. It is possible to change on time, is very essential. the message handler core as well as certain other cores Figure 3 illustrates the communication model adopted by having an interface card connected to the module and by TTCAN. The matrix cycle essentially describes the an external CPU. transmission schedule of the entire network. It defines In summary, with the addition of a session layer to which node should be transmitting at any given instant the protocol stack of a CAN network, it is possible to of time. In this way, each node knows when exactly it have a time-triggered model of the CAN that lends is supposed to transmit. The matrix cycle is repeated flexibility as well as a deterministic behaviour to the network. over time. Each basic cycle consists of several exclusive This fact will be exploited by systems such as x-byover and/or arbitrating and free windows. This structure is wire in the future. Slowly, such networks would be added highly column oriented. Each column in a matrix cycle to the existing mechanical and/or hydraulic back-ups to is known as a transmission column. All the windows ensure the complete safety of the operating control belonging to a single transmission column are equal in system. The advantage of using a time-triggered archi- tecture is that the system would not be overloaded with events. This would take additional circuitry to resolve pending scenarios. 4 SAE J1850 PROTOCOL The SAE J1850 protocol has been defined by SAE to be a class-b protocol, whose definition is given as A system whereby data, e.g., parametric data, is transferred between nodes to eliminate redundant sensors and other system elements. The nodes in this form of a multiplex Fig. 3 Matrix cycle in TTCAN system typically already existed as a stand-alone module

5 ELECTRONIC CONTROL UNITS FOR AUTOMOTIVE ELECTRICAL POWER SYSTEMS 1221 in a conventionally wired vehicle [11]. Class-B networks the bus. The priority of the messages is indicated by bits are essentially used for communicating between modules in the beginning of the messages. A CRC is implemented supporting a data transfer rate of around 100 kbps. in the protocol for error correction [12]. Though this data rate is insufficient for real-time safety Once the transmitting node has finished transmission, critical processes, class-b networks are ideal for nonaccess the nodes that had lost arbitration try again to gain real-time control processes and communication. to the bus and the whole arbitration procedure is The SAE J1850 protocol is based on a bus level repeated again. Invariably, it is seen that nodes that have topology wherein the bus is a masterless peer-to-peer more active symbols in the starting of the message com- protocol. Just like the CAN network, messages are pared to others gain access to the bus earlier than the broadcast in the sense that whatever messages have to other messages. However, it must be stressed that, due be sent are sent in the form of frames to all the nodes, to the nature of the bus access, it is impossible to specify irrespective of the desired destination. instants of time at which certain nodes can transmit or This protocol supports two modes. One is a 41.6 kbps receive, very much unlike the TTCAN presented before. pulse-width modulated mode supporting a dual wire and The entire SAE J1850 protocol is implemented on a the other is a 10.4 kbps variable pulse-width mode supdeveloped single silicon chip. For example, the data link controller porting a single-wire configuration. Data can be sent on by Delco Electronics Corporation for General the bus by way of either time division multiplexing or Motors [13]. This chip has an 11-byte transmit first-in- frequency division multiplexing. In time division multibuffer as well as an 8-bit parallel data interface. It also first-out ( FIFO) buffer and a 20-byte receive FIFO plexing, messages are sent on the same bus at different instances of time whereas in frequency division multi- has a transceiver built on. This transceiver provides a plexing two or more messages are sent at the same time. 7 V waveform for transmission on the bus. Wave shaping Data on the bus is transmitted by changing the amount is done to reduce RFI emissions. Functions, such as of time each transition lasts on the bus. For example, filtering, timing, and symbol definitions, are performed according to VPW J1850 protocol definition, a 1 bit is by the logic section. The logic section includes host a bus driven at a high potential (active) state for 64 ms interface, error detection, and correction codes, and all kinds of logic. A diagram of the DLC chip is shown or low potential (passive) state for 128 ms. Alternatively, in Fig. 4. a 0 bit is defined as a bus in high potential (active) state The total length of the network wiring is restricted to for 128 ms or low potential (passive) state for 64 ms. A 40 m. This included 35 m of on-vehicle networking and high potential is usually any voltage from 4.25 V to 20 V. 5 m of off-vehicle networking. The wiring itself is a single A low potential is any voltage from ground voltage to wire in the 10.4 kbps case and double wires in the case of 3.5 V. the 41.6 kbps model. The maximum number of modules The transitions between the voltage levels are achieved or nodes, including off-vehicle equipment, is 32. This by using transistors. When the bus is in idle state it is at assumes that a single node is seen as a unit load, which low or ground potential. When a node wants to transmit, corresponds to 10.6 K V and 470 pf. Off-vehicle miniit uses the transistor to push the bus to a high state, mum resistance should be 10.6 K V and the maximum thereby overriding any node wanting to pull the bus capacitance should not exceed 500 pf. Another major to a low state. Here, nodes that wish to transmit first specification is that a 0 bit symbol is the dominant sense the bus for a fixed amount of time. If they sense symbol and the 1 bit symbol is the recessive symbol. the bus to be idle, they start transmitting. If they sense the The pulses on the 10.4 kbps model are all wave-shaped. bus is busy, they wait for a certain amount of time before The advantage of using this shape is that it rounds off trying to start transmission once again. Let us assume the sharp edges of the bus transition, thereby removing that a node wishing to transmit transmits a passive unwanted high frequency components. Table 1 shows symbol. If there is no challenge, it goes ahead and trans- the VPW DC parameters. mits the message. It has to be noted here that even during The advantage of the SAE J1850 is that it is an transmission each bit is checked to makes sure that there open architecture. There are no stringent restrictions in is no other node with higher priority waiting to transmit. its implementation and development, and hence it gives This is called bit-by-bit arbitration. This process is engineers and designers sufficient leeway in developcarried out until only one bit is left. However, before ing applications. It is mainly used for diagnostic and transmission, if there is a node transmitting an active data-logging functions. symbol, the node with the passive symbol looses control of the bus and starts functioning as a receiver. The node with the passive symbol is then said to have lost 5 IEEE 1394 PROTOCOL arbitration. Thus, we see why the arbitration scheme used is called carrier sense multiple access/collision The IEEE 1394 is an up and coming bus standard that resolution (CSMA/CR). It is called multiple access because can provide a high rate of data communication between it is possible for multiple nodes to gain equal access to electronic devices. The bandwidth offered by an IEEE

6 1222 A SHRINATH AND A EMADI Fig. 4 DLC chip diagram Table 1 VPW DC parameters Parameters Symbol Minimum Type Maximum Unit Input high voltage V ih Volts Input low voltage V il 3.5 Volts Output high voltage V oh Volts Output low voltage V ol Volts Absolute ground offset voltage V go Volts Network resistance R load Ohms Network capacitance C load PF Network time constant T load 5.2 Ms Signal transition time T t 18.0 Ms Node resistance (unit load) R ul Ohms Node capacitance (unit load) C ul 470 PF Node leakage current I leak 10 MA 1394 serial bus is usually from 2i to Mbps, with are translated into data at the other end. The link layer i taking on the values of (i=0, 1, 2, 3, 4, 5) [14]. It has is associated with addressing as well as data checking. a 64-bit addressing system, out of which the 16 most It is in this layer that error detection takes place. The significant bits make up the node ID [15]. Out of these transaction layer basically provides the set-up mech- 16 bits, ten bits are used for bus Ids, which give rise to anisms for communication to take place. The serial 210 1=1023 buses and the remaining six bits giving management bus has management functions associated rise to 26 1=63 nodes. A typical application for this with it. It is linked to all three layers. Finally, the appli- protocol is for in-car infotainment purposes. The proapplications cation layer is the layer in which codes for specific tocol stack implemented in the IEEE 1394 is shown in are written. Fig. 5. The 1394 protocol supports a variety of features, such Just like the OSI model that is used widely in the as high data rates and plug-and-play compatibility, as TCP/IP protocol each level in the IEEE 1394 stack has well as support for asynchronous and isosynchronous certain functions assigned to it. The physical layer is in data types [16]. Asynchronous data transfer is mainly charge of sending electrical impulses on the bus, which used for guaranteed data delivery, which some protocols such as the SAE J1850 protocols, offer (the in-frame response). Messages are short and are mainly used for control and setup purposes. Isosynchronous data transfer is used for mass transfer of data. However, reliability is not a major criterion in this case. Constant speed and bandwidth are the major factors taken into consideration for this kind of data transfer. Messages are sent in the form of packets, which are sent out every 125 ms. A typical application is video streaming. The IEEE 1394 protocol provides for automatic configuration of the tree topology implemented. The three basic steps are: bus initialization, tree identification and Fig. 5 IEEE 1394 protocol stack self-identification. In the case of dynamically changing

7 ELECTRONIC CONTROL UNITS FOR AUTOMOTIVE ELECTRICAL POWER SYSTEMS 1223 centralized approach, a single node handles all the tasks. In the decentralized approach, all the administrative tasks are distributed throughout all the nodes. A MOST network is essentially made up of MOST interconnect, MOST system services, and MOST devices. The MOST interconnect is essentially concerned with the establishment of connections between the devices at the start-up phase. It is also during this phase that device addresses are distributed throughout all the devices. Synchronization between all the nodes by means of a bit pulse is achieved during this phase. The MOST system services are comprised of low-level system services. Functions such as data routing, channel allocation, fault detection, or delay detection are per- formed by the low-level system services. The application socket and the basic layer system services make up the MOSTNetservices, which are basically concerned with the transmission and reception of different kinds of data, as well as providing standard interfaces to access various network management functions. MOST devices can vary from simple displays to complex applications. They have the capability of bandwidth allocation, as well as packet and control data capacity handling. Figure 6 illustrates the implementation of a MOST device in a typical configuration. The data types supported by MOST are control data and bursty data. All the transactions taking place on the MOST network involve the generation of a synchronous frame by a frame generator or timing master. Each node has an internal timing device that locks onto this signal by means of a phase-locked loop (PLL). By locking onto this synchronization frame, all the nodes are synchronized to the timing master. MOST networks are designed to use fibre optical cables as the medium of transmission. Therefore, the network is designed to accommodate all the possible complex topologies that can implement applications based on fibre optical cables. This provides enhanced capacity, as well as security features. It also enables the MOST network to support the plug-and-play mode of operation. The signal lines that compose the physical structure are RX-receive data and TX-transmit data. Other optional lines are power, ground, and wake-up. As the name suggests, the wake-up line is used for waking-up the target device. In this network, each node IDs, the devices must implement a group of registers defined by the control and status register (CSR) architecture specification. The devices themselves are generally of two types: low-cost devices and high-end devices [14]. The low-cost devices are usually the devices at the user end of the system (e.g. air-conditioning, music system, and motor electronics). These devices usually contain few codes and a microcontroller. The high-end devices are usually more powerful, having a powerful processor and universal user interface (e.g. touch screen or display board replete with keyboard and mouse). Such powerful devices have uses in the field of GPS and navigation. Another important characteristic of the IEEE 1394 protocol is that the management functions are distributed. This means that the nodes have no master and each node operates independently. Therefore, even if one part of the network ceased to function for some unforeseen reason, the rest of the network would continue to function irrespective of the failure of a part of the network. Main applications include the avionics industry to handle CNI waveforms, industrial cameras for industrial inspection systems for quality control, and multimedia applications in automobiles. 6 MEDIA ORIENTED SYSTEM TRANSPORT (MOST) The development of the MOST network [16] stemmed from the need for a peer peer network that had high data speeds, but was also cost effective. The MOST network was essentially designed to be used for multimedia applications in the automotive environment; however, due to its efficiency, it is increasingly being used for home networks too. Also, the implementation of optical fibres at its physical layer ensures high reliability under adverse circumstances. Some of the key features of the MOST networks are ease of use, cost effectiveness, availability of synchronous and asynchronous data transfer mode, flexibility, and a wide range of applications. As pointed out before, the MOST network is a peer peer network. The point-to-point network connection can be established using either of the ring, star, or daisychain topologies. Two approaches can be used for administrative tasks: centralized and decentralized. In the Fig. 6 MOST device in a typical configuration

8 1224 A SHRINATH AND A EMADI has a set of functions that are to be carried out. The most important ones are: synchronization, managing data flow, decoding addresses, detection of system start-up, and power management. These are the functions that are implemented in each node. However, the software too has many functions that are important. Some of them are: physical addressing, channel allocation, control, data transfer, packet data transfer, and system monitoring. Since this network can support real-time data transmission, it can be used for networking CD audio drivers, set-top boxes, and TV sets. Current MOST technology can support over 12 uncompressed stereo CD quality audio channels at the same time or 12 MPEG1+ audio Fig. 7 An implementation of x-by-wire system for steering channels or several MPEG2+ audio channels. It can also be used for networking multimedia at homes, for example multiroom audio and video devices. and braking systems In the automotive field, MOST is applied for The entire architecture of the x-by-wire system is based implementing infotainment networks. Data, such as door on a time-triggered approach. Here, all the activities are status or illumination property signals, are transmitted carried out at specific instants of time. Time is divided to the body control module using MOST technology. into slots of equal duration. For each time slot, there is 7 X-BY-WIRE a specific task assigned. All the nodes are synchronized to one global sense of time. This is achieved by the transmission of a synchronization pulse by a master node. This pulse is periodic in nature. The automobile industry is always concerned with safetyrelated At specific instants of time, all the sensors capture issues. It is always trying to develop methods their measurements, which are, in turn, transmitted on that keep increasing safety standards, for example, the bus. These measurements are used to update certain intelligent driver assistance. However, such systems need variables. The new values overwrite the old ones. These to be computer-controlled to deliver maximum efficiency. new values are then used for control applications. With this comes the need to replace all the mechanical A main feature of the x-by-wire system is what is known or hydraulic backup with electric/electronic components. as composability. It means that whatever properties This can only be done when it has been ascertained are exhibited by subsystems, the same properties are that the systems that are replacing the mechanical or exhibited by when the various subsystems are integrated hydraulic backups are very safe. Such systems are known to offer some function or application. Another important as x-by-wire [17] systems. A consortium comprising feature is the ability of the system to handle errors that Daimler-Benz, Centro Ricerche Fiat, Ford Europe, are generated when nodes talk during a slot that is not Volvo, Robert Bosch, Mecel, Magneti Marelli, the allotted to them. This node is known as a babbling University of Chalmers, and the Vienna Institute of idiot. This error can be avoided because, in a timetriggered Technology has carried out work in this field. Examples system, it is known beforehand which node is of this kind of system are steer-by-wire and brake-by- supposed to transmit at which time interval. wire systems. Figure 7 illustrates a typical application of The bus access is controlled by a TDMA scheme. an x-by-wire system. Broadcast topology is used. As mentioned before, The system is comprised of actuators and sensors connected information regarding which node can transmit at which to the ECUs. The sensors and actuators are used interval is stored in the memory. This information is for taking measurements, which are, in turn, fed to the stored in every node locally. Every node is supposed to ECUs for driver feedback. Based on the measurements follow this timetable. The protocol used is the timetriggered taken, the driver can make suitable modifications, which protocol/class-c ( TTP/C). are then relayed to the actuators for implementation. The node architecture is based on what is known as It must be ensured that the measurements taken are a fail-silent mode of operation. This means that if a always accurate. If the readings taken by the sensors are node is detected as having an error, it is asked to stop not accurate or the response of the driver given to the transmission. This is done so that the erroneous node actuator is not accurately interpreted, it could lead to does not interfere with the normal performance of the very serious losses. Another problem with implementing network. Sometimes, two nodes are used as a pair for fault tolerance. If one node has gone silent, then the other node can continue to perform the task in an error- free way. This technique has many distinct advantages. this technology for mass production is its economic feasibility. In addition, associated with this technology are problems such as reliability and maintainability.

9 ELECTRONIC CONTROL UNITS FOR AUTOMOTIVE ELECTRICAL POWER SYSTEMS 1225 First of all, the need for complex circuitry to implement other fault-tolerant mechanisms is avoided. Second, combined with software fault detection, the fail-silent architecture would be able to cover the entire range of possible errors comprehensively. Third, this mechanism can be tested to test its full functionality. Each ECU, as mentioned before, is made up of some sensors and actuators. It must be ascertained that in a node comprising two ECUs there should not be any disparity in the decision finally taken. This may mean that each ECU exchanges sensor values with the others on the common bus, so that a common decision is reached. An alternative to using the common system bus is using a sensor bus. Also, if digital sensors are used, each ECU will have the same values. If the two ECUs are located far way from each other, then the sensors are connected to the closest ECU. If, on the other hand, they are situated close to each other, then the sensors may be connected to the ECUs by means of a sensor bus as mentioned above. Figure 7 demonstrates the concept of implementation of a X-by-wire system in a vehicle. The software model consists of two layers: the system software layer and the application software layer. The system software layer is used to provide services to the application software layer. Services such as fault tolerance and detection are implemented in the system software level. The language used to write codes is the ANSI C subset with certain modifications, such as exclusion of certain error-prone constructors and inclusion of exception handling. Efficient design, higher fault tolerance, ability to be tested, and easy-tosynchronize are the main advantages. The x-by-wire system is being developed by the European Consortium to provide a standard that is accepted all over Europe for the development of intelligent driving aids for drivers. 8 LOCAL INTERCONNECT NETWORK Local interconnect network (LIN) [18] is a communication protocol that has been established for automobiles. This protocol is based on the SCI (UART) data format with a single master/multiple slave architecture. A consortium formed in 1998, comprising Audi, BMW, Daimler Chrysler, Volvo, Volkswagen, VCT, and Motorola worked on establishing a specification for this protocol. The development of LIN was based on the necessity for a communication protocol that was very cost effective and not only addressed the issue of the specification for communication, but also other issues like signal transmission, programming, and interconnection of nodes. It basically takes an all-round approach for the development and consolidation of an automotive protocol. The basic advantage that LIN enjoys is that it is very economical compared to protocols such as CAN. However, this advantage is negated by its inherent limitations, such as low bandwidth and performance, and the single master topology of the network. There are quite a few criteria that have to be taken into consideration when a network is designed. Factors such as bandwidth, security, latency, electromagnetic interference, fault tolerance, and cost should be balanced in order to design a network to suit the requirements. There are, basically, two different approaches that could be taken to design a network. One approach is to divide the sensors and actuators in a zonal manner, connected to a central ECU. Various main ECUs are connected to each other by CAN links. This extensive use of CAN is to enable the usage of high bandwidth for signal exchanges. The other approach is to totally abolish the zonal concept. Here, all the actuators and sensors are connected to the central ECU by means of LIN links. This has the advantage of being scalable. No major changes have to be done to the network in order to accommodate additional nodes. As mentioned before, LIN is based on the SCI (UART) byte word interface. The network has a single master/multiple slave topology [19]. All the slaves have the job of transmitting and receiving. The master node, apart from the task of transmission and reception, has the additional task of maintaining the synchronization in the network. This is done by means of a message header, which consists of the synchronization break, synchronization byte, and a message identifier. The message identifier is used by the nodes to identify the messages meant for them. Each identifier is unique to the node. This way the node knows the messages meant for it. It has to be remembered that the message identifier indicates the content of the message and not the destination. Once the node has received the message meant for it, it then sends back a response that contains data with the size ranging from two, four, or eight bytes of data with one checksum byte. One message frame consists of the header and the response parts. Each message frame is made up of a byte field, which has ten bits. These ten bits include a dominant start bit, eight bits of data, and a recessive stop bit. The message frame consists of the header sent by the master node and the response sent by the slave node. The header sent by the master node consists of synchronization break, synchronization field, and the identifier field. Each message frame is built on the 8N1-coding scheme. The synchronization break must be a minimum of 13 bits in length to ensure proper synchronization. The synchronization field is a string of bits with an equivalent hexadecimal value of With this type of syn- chronization pattern, it is possible for nodes not equipped with quartz stabilizers to resynchronize them- selves. There are basically two levels of synchronization that are defined for the LIN network: unsynchronized, in which the slave clock time differs from the master

10 1226 A SHRINATH AND A EMADI clock time by less then ±15 per cent and synchronized, Bluetooth is implemented in devices by way of small in which the slave clock time differs from the master chips. It uses a shortwave radio signal that is always on clock time by less than ±2 per cent. for signal transmission purposes. It creates what is There are four identifiers reserved for special purposes. known in the technical terms as an ad hoc network [21]. Out of the four identifiers, two are used for uploading The participants in this ad hoc network are the slaves and downloading purposes. The only way they differ from and the masters present in the entire network. Each the normal data frame is that in these frames, instead of master and slaves form what is known as a piconet. data, there are user-defined command messages. The This, basically, represents an area of active participation other two special identifiers are used for ensuring upward by one master and one or more slaves. A slave can compatibility with the future versions of the LIN protocol. participate in any number of piconets, but a master can These identifiers are called extended identifiers. belong to only one piconet. The topology is also called There are three types of communication mode that are the scatternet topology. Each piconet has its boundary. supported: master to slave or multiple slaves, slave to There is at least one slave acting as a bridge between master, and slave to slave: the slaves can talk to each two piconets. Each piconet has its own characteristic other without routing the transmission through the transmission frequency, also, since each piconet uses its master. This is illustrated in Fig. 8. own transmission frequency, each piconet is uniquely A typical LIN network has a single master with as identified by the hopping pattern used for transmission. many as 2 10 nodes contained in the network. Typical The master of that particular piconet determines the data speeds range from 2.4 kbps to 19.6 kbps. The volt- hopping pattern. age level that can be supported by the buses is around The spectrum of Bluetooth lies in the unlicensed 13.5 V. Each data frame contains 2 8 bytes of data with industrial, scientific, and medical ( ISM) band. The frequency a six-bit identifier. Single-wire transmission, low cost of ranges from 2.4 GHz to GHz. A number implementation, and no resonators are the advantages. of multiple access schemes are used. The most widely Low bandwidth and a single-bus access scheme are the used schemes are the time division multiple access main disadvantages. LIN is usually used for door and ( TDMA), frequency division multiple access ( FDMA), roof control, as well as for the steering wheel and column. and code division multiple access (CDMA). A total of Other implementations include smart wiper motor, 79 hop frequencies have been defined spaced 1 MHz sensors and switch panels, seat control, and heating. from each other in the United States. However, in 9 BLUETOOTH France and Spain, instead of 79, there have been only 23 hop carriers defined. This protocol supports both the synchronous connection oriented (SCO) link as well as the asynchronous connection-less (ACL) link. Bluetooth [20] is a wireless communication standard The Bluetooth baseband model is the intelligent that has been developed for networking small devices core of communication. The functionality of the module such as printers, palm pilots, and mobile phones. It was depends on the kind of the application. Simple appli- initially conceived by Ericsson in Since then com- cations can be run on the baseband processor, whereas panies such as Nokia, IBM, Intel, and Toshiba have complex applications are run on the host processor. joined it in its research efforts and in 1998 formed what Therefore, the module should be very flexible in terms is known as the Bluetooth Special Interest Group (SIG). of load sharing; it must be able to accept any load at Currently, about 2000 companies are part of the SIG any point of time. Efficient usage of resources is a prime and are developing Bluetooth-related products. issue to be considered when designing a module. There are several modules that have been developed. But, their usage has been hampered by the fact that they are large due to huge hardware. All the functions are performed by this hardware block instead of the microcontroller. This leads to a waste of the microcontroller. Therefore, the modules that existed so far were inefficient. However, to increase the efficiency of the module it is possible to transfer a lot of functionality to the microcontroller, with the hardware blocks implementing only essential hardware functions. The basic block diagram of a module is shown Fig. 9. The microcontroller is used for controlling other units, as well as controlling some link control protocols. The baseband unit is entrusted with the task of encoding and decoding bit streams, as well as control of the RF Fig. 8 Data communication in LIN module. The USB and UART blocks are used for

11 ELECTRONIC CONTROL UNITS FOR AUTOMOTIVE ELECTRICAL POWER SYSTEMS 1227 Fig. 9 Bluetooth baseband module implementing the host control interface ( HCI) and the physical layer. The audio CODEC, as the name suggests, is used for audio coding and decoding as well as supporting both the A-law and the m-law modulated schemes. 9.1 Bluetooth in vehicles it is very difficult to integrate different applications on a common network. With the employment of Bluetooth, it is possible to integrate various networks on the backbone that Bluetooth offers. This can lead to exciting new applications. A mobile environment like a car can then be turned into virtually a mobile office station where office work, such as s or file transfer, can take place; a mini theatre where a mini screen in the car can play a movie being broadcast with a touch of a button on a keypad, which acts like a wireless remote control; or it could be a simple application like browsing the internet and booking an airline ticket online. The possibilities are tremendous. What remains to be seen is the efficiency of this technology. With so many devices linked by radio and so many applications, the question of efficient bandwidth utilization and security issues, to name but two, have to be answered. It must be mentioned that Bluetooth would initially be applied to safety-uncritical applications in and around the car, for example, communication and infotainment applications, or applications that involve exchange of pure car-specific information (control and status data of the body electronics). Inside the car, Bluetooth allows the wireless connection of various car-embedded electronic devices, such as control panels and headsets, with other car-embedded electronic devices. However, beyond replacing wires between in-car devices, the scope of Bluetooth in the automotive industry is to ensure interoperability between wireless devices (e.g. mobile phones and PDAs), and car-embedded devices (e.g. car navigation systems, car radio, and car-embedded phones). The main principle here is to allow devices to cooperate and share resources and during almost the entire time, access the car-embedded system through mobile devices. This is implemented by the way of a central control unit (CCU ) [23]. Apart from controlling various devices in the car, Bluetooth can be used in car car communication, as well as in other Bluetooth-enabled devices near the car s environment. The exchange of data between the car and stationary facilities provides a base for various special application scenarios, e.g., access of traffic information or e-commerce applications at designated access points, or download of software for car electronic systems for service stations. Listed below are some of the appli- Automotive manufacturers have to deal with problems arising from the use of wiring of electronic parts, including networking, power efficiency, weight factor, and shrinking layout space, as well as assembly-line difficulties to name but a few. There is a wide range of automotive networks, reflecting defined, functional, and economic niches. High bandwidth networks are used for vehicle multimedia applications, where cost is not critical. Reliable, responsive networks are critical for real-time control applications, such as powertrain control and vehicle dynamics. Not only can Bluetooth be used for control applications, it can also be used to network multimedia devices. Vendors have envisioned a scenario where it is possible to obtain data from a PDA, as well as control multimedia operations in a car via a Bluetoothenabled cellular phone [22]. Applications running on voice recognition can also be added on for applications such as banking or shopping online, which can then be implemented using Bluetooth technology. Therefore, using this technology the car can be made into a virtually moving office. As mentioned before, there are many applications that can be supported in a vehicle. Each application demands a different kind of a network for cations that are envisaged for Bluetooth in the near the devices implementing that particular application. future [24]. The reason is that every application may have varying demands of bandwidth, security, and reliability. Anytime information. With this kind of information Connectivity and connection with the world outside the car enable new applications. Vehicles can make use of various communication networks available outside the car from communication services to applications inside the car. It is at this conjuncture that Bluetooth comes into the picture. The problem with the existing networks for a mobile environment like a car is that there are so many networks involved, each one having its own data transmission rate and characteristics, that service, users receive data via broadcast or multicast. The main advantage of these services is their always on nature, such that services can be accessed without any setup effort. Moreover, data carousels allow starting reception at any time. A certain casting service may be offered in large areas or in limited areas only, according to the coverage of the broadcast program. In this way, it is subject to variation along the travelling route of the automobile. The received information can be displayed