Multiple single-chip microcomputer approach to fire detection and ...

Multiple single-chip microcomputer approach to fire detection and monitoring system A.J. AI-Khalili, MSc, PhD D. AI-Khalili, MSc, PhD M.S. Khassem, MSc

Indexing t e r m : Hazards, Design, Plant condition monitoring

Abstract: A complete system for fire detection and alarm monitoring has been proposed for complex plants. The system uses multiple single chip architecture attached to a party line. The control algorithm is based on a two-level hierarchy of decision making, thus the complexity is distributed. A complete circuit diagram is given for the local and the central station with requirements for the software structure. The design is kept in general form such that it can be adapted to a multitude of plant configurations. It is particularly shown how new developments in technology, especially CMOS single chip devices, are incorporated in the system design to reduce the complexity of the overall hardware, e.g. by decomposing the system such that lower levels of hierarchy are able to have some autonomy in decision making, and thus a more complex decision is solved in a simple distributed method. 1

Introduction

Regulatory requirements for most high risk plants and buildings mandate the installation of fire detection and warning systems for all sensitive areas of the plant or the building [l, 3, 41. Most fire codes state the requirement for monitoring and control specifically related to a type of a plant or building such as chemical plants, petroleum, nuclear plants, residential high-rises etc. A general conclusion of these codes can be specified as the following requirements : (a) the source of all detector signals should be exactly identifiable by the central station (b) an extra path of communication between the central station and all local controllers (c) direct means of control of alarm and central equipment by the central station (d) means of communication between the central station and the fire department (e)availability of emergency power supply. The codes usually also specify the types and frequency of tests for all equipment. A fire detection and alarm system is a combination of devices designed to signal an alarm in case of a fire. The ~

Paper 58256 (El, ElO), first received 24th March and in revised form 10th November 1987 Dr. A.J. AI-Khalili is with the Department of Electrical Engineering, Concordia University, Montreal, Canada H3G 1M8. Dr. D . AI-Khalili is with the Royal Military College, Kingston, Ontario, Canada K7L 2W3 IEE PROCEEDINGS, Vol. 135, P t . G , N o . I , FEBRUARY 1988

system may also accomplish fan control, fire door hold or release, elevator recall, emergency lighting control and other emergency functions. These additional functions supplement the basic system which consists of detection and alarm devices and central control unit. Technology has an influence on system architecture. When technology changes, the architecture has to be revised to take advantage of these changes. In recent years, VLSI technology has been advancing at an exponential rate. First NMOS and, in the last year or two, CMOS chips have been produced with the same packing density with more gates per chip yet at a lower power consumption than NMOS. Surely this change in technology must affect our design of hardware at both the chip and the system level. At the chip level, single chips are now being produced which are equivalent to board levels of only the previous year or two. These chips have microprocessor, memory in RAM and ROM, IO Ports both serial and parallel, A/D timer, flags and other functions on chip. At the system level, the new chips make new architectures possible. The objective of this paper is to show how technology can influence system architecture in the field of fire control. The new high density single chip microcontrollers are incorporated in the design of a large scale system and yet we obtain a smaller system with a better performance. In terms of fire detection and alarm monitoring, this is reflected directly in the local station hardware, because of their remoteness and power supply requirements. A complete local station can be designed around a single CMOS chip with power consumption of a few mW depending on system operation. This approach reduces the cost and complexity of design, implementation and maintenance and provides easily expandable and portable design. This implementation was not possible with old technology. Most of fire detection/monitoring systems available [2, 3, 51 are tailored towards a specific application and lack the use of recent advances in CMOS VLSI technology. In this study, we develop a fire detection/monitoring system which is general in concept, readily implementable in a multitude of applications for early detection of a fire before it becomes critical, for equipment and evacuation of personnel. Here, we propose a central control and distributed control/detection/monitoring with adequate communication, where use is made of single-chip microcontrollers in the local stations, thus improving controllability and observability of the monitoring process. 2

Detection and alarm devices

A basic fire detection system consists of two parts, detection and annunciation. An automatic detection device, 1

such as a heat, smoke or flame detector, ultraviolet or infrared detectors or flame flicker, is based on detecting the byproduct of a combustion. Smoke detectors, of both ionisation and optical types, are the most commonly used detector devices. When a typical detector of this type enters the alarm state its current consumption increases from the pA to the mA range (say, from a mere 15pA in the dormant mode to 60 mA) in the active mode. In many detectors the detector output voltage is well defined under various operating conditions, such as those given in Table 1. The more sensitive the detector, the

radiation are used when flammable liquids are being handled, heat detectors are used for fire suppression or extinguishing systems. In general, life and property protection have different approaches. Alarm devices, apart from the usual audible or visible alarms, may incorporate solid state sound reproduction and emergency voice communication or printers that record time, date, location and other information required by the standard code of practice for fire protection for complex plants. Heavisid [4] has an excellent review of all types of detectors and extinguisher systems.

Table 1 : Typical detector voltage output levels

2.1 Control philosophy and division of labour Our control philosophy is implemented hierarchically. Three levels of system hierarchy are implemented, with two levels of decision making. There is no communication between equipment on the same level. Interaction between levels occurs by upwards transfer of information regarding the status of the subsystems and downwards transfer of commands. This is shown in Fig. 1 where at level 1 is the central station microcomputer and is the ultimate decision maker (when not in manual mode). At level 2 are the local controllers, which reside in the local stations. At level 3 are the actual detectors and actuators. A manual mode of operation is provided at all levels. Information regarding the status of all detectors is transmitted on a per area basis to the local controllers. Their information is condensed and transmitted upward to the

Voltage level

Detector condition

22 v 13 V-22 V 1 V-13 V 1 v

Detector is in open circuit condition detector is normal detector is in an alarm state detector is in short circuit condition

more susceptible it is to false alarms. In order to control the detector precisely, either of the following methods is used: a coincidence technique can be built into the detector, or a filtering technique such that a logic circuit becomes active only if x alarms are detected within a time period T . The detection technique depends greatly on the location and plant being protected; smoke detectors are used for sleeping areas, infrared or ultraviolet

1 L

.- --..... ......

rn

ACTION STATION

WARNING

~

LEVEL 3

2

Fig. 1

Control philosophy

IEE PROCEEDINGS, Vol. 135, Pt. G, N o . I , FEBRUARY 1988

central microcomputer. Transfer of status is always unidirectional and upwards. Transfer of commands is always unidirectional and downwards, with expansion at the local control level. This approach preserves the strict rules of the hierarchy for exact monitoring detection and alarm systems associated with high risk plants. The classification of the two layers of controls is based upon layers of decision making, with respect to the facts that (a) when the decision time comes, the making and implementation of a decision cannot be postponed (b)the decisions have uncertainty

(c) it will isolate local decisions (e.g. locally we might have an alarm although there may be a fault with the system) 3

General hardware

Fig. 2 depicts our design in the simplest of forms. The system uses an open party line approach with four conductor cables going in a loop shared by all the remote devices and the control panel. This approach is simple in concept and is economically feasible. However, one major disadvantage is the dependency on a single cable

H I

FROM INPUT DEVICES

I .t

TO OUTPUT

REMOTE MC

REMOTE MC

CENTRAL MICROCOMPUTER

TWISTED PAIR FOR SIGNAL

Fig. 2

System schematic open line connection

I

HIGH CURRENT OUTPUT

1

HIGH CURRENT OUTPUT

HIGH OUTPUT OUTPUT 3

2

I

SUPERVISED OUTPUT

DETECTOR 1

I I

I

L I

41

: I

I I

SINGLE CHIP MICROCOMPUTER

I I I II

I I

I I I I

LINE DRIVERS AND RECEIVERS

SUPERVISED INPUT UNSUPERVISED INPUT UNSUPERVISED OUTPUT

Fig. 3

IEE PROCEEDINGS, Vol. 135, Pt. G , N o . I , FEBRUARY I988

Block diagram ofremote station 3

t 16 v

A D VREF

nicn

MODE A

A D V REF LOW

I

PORT A

SWITCH RESET LOCAL

I

I

CND

-

TO DET. 4

2.4576 PCO

EXTAL RX

TX

-

A

*

TO DET. 3 TODET z

Y = PULL-UP RESISTORS UNSUPERVISED INPUTS

RECEIVER

(C

TYPICAL DETECTOR CONNECTION

TO DET. 5

X-TAL

33

-

NON SUPERVISED OUTPUTS

DETECTOR ENCLOSURE

DRIVER

5" 100 MA

10 MF

2

A

1: M S T E D PAIR

+ 26V

I1

GND

Fig. 4

4

Remote station circuit diagram

IEE PROCEEDINGS, Vol. 135, Pt. G , No. I , FEBRUARY 1988

for power and signalling. In cases where reliability is of extreme importance, two or even three cables taking different routes throughout the system may be connected in parallel. Fig. 3 gives the driver circuitry required to derive an expandable bus. This design takes advantage of recent advances in the single chip microcomputer technology to reduce the interface between the central station and the local stations. 3. I Central control task A central unit provides a centralised point to monitor and control the system activities. In the system to be described the central control unit serves a fivefold purpose. (i) It receives information from the local stations and operates the alarms and other output devices. (ii) It notifies the operator in case of system malfunction. (iii) It provides an overall system control manual and automatic. (iu)It provides a system test point of local stations and itself. ( u ) It provides a central point for observation, learning and adaptation.

3.2 Local stations The local stations can take local decisions regarding recognition of a risk situation, and act independently on local affairs. In this technique we depend on ‘load-type coordination’, e.g. the lower level units recognise the existence of other decision units on the same level; the central or the top level provides the lower units with a model of the relationship between its action and the response of the system. It is evident that a powerful machine is required at this stage so that all the required functions can be implemented. The availability of the new generation of microchips makes this architecture a feasible solution. A single chip microcomputer was chosen over discrete digital and analogue devices to interface to the field devices and to the central microcomputer. This is the FROM INPUT DEVICES

1

4

System implementation

The local station: Fig. 3 is the block diagram of the circuit used to utilise the MC68HCllA4 as a remote fire detecting circuit while Fig. 4 illustrates the same circuit in an expanded form. It can be seen that the single microcontroller can be used to monitor more than one detector, thus reducing system cost. The loop power supply, which is usually between 28 and 26 V, is further regulated by a 5 V 100 mA monolithic low power voltage regulator to supply power to the microcontroller. The onboard oscillator, coupled with an external crystal of 2.4576 MHz, supplies the microcontroller with its timing signal which is divided internally by four to yield a processor frequency of 614.4 kHz, which is an even multiple of the RS 232 [7] baud rate generator. In this Section the term ‘supervised input or output’ will be used to mean that the function in question is monitored for open- and short-circuit conditions in addition to its other normal functions. More information can be found in Reference 9.

TO OUTPUT DEVICES

REMOTE

REMOTE M.C.

-

M .C. 1

-

L

2

REMOTE M.C. 3

+ 26V

A

REMOTE

M.C. N

-

b

I

-

main reason that previously this approach was not feasible. In selecting the microcomputer for the local stations, the criterion was the requirement for a chip which contains the most integration of the analogue and digital ports required for the interface and the utilisation of CMOS technology owing to remoteness of the local stations. The choice was the Motorola 68HC11A4, for the following reasons: ( a ) It is CMOS technology; this reduces power consumption. (b)It has a UART on board; this facilitates serial communication. (e) It has an a/d converter on board; this eliminates an external A/D. ( d ) It has 4K of ROM, 256 bytes of RAM, 512 bytes of EERROM with 40 1/0 lines and a 16 bit timer; this satisfied all our memory and 1 / 0 requirements at the local station side.

-

-

A

A

A

POWER LINES

I

1

--

CENTRAL MICROCOMPUTER

f-DIRECTION OF TRANSMISSION

IEE PROCEEDINGS, Vol. 135, Pt. G , N o . I , FEBRUARY 1988

Fig. 5

Closed loop connections

5

4.1 Supervised detector connections

4.2 Communications channels

Power is applied to five detectors via a series pass transistor, the function of which is to interrupt the flow of power to the detectors upon command from the microcontroller in order to unlatch the output SCR from its alarm condition. The maximum detector current consumption is 65 mA. R1 serves as a short-circuit current limiting device in case of a short-circuited detector. The 10 KO end of line resistor (EOL) helps the supervising circuit detect the presence of the detector since its normal state input impedance is very high, making it very dificult to distinguish between an open circuit condition and a normal condition without the EOL resistor. The positive detector terminal A is connected to the microcontroller A/D converter circuit via an R/5R divider circuit, the function of which is to reduce the detector output voltage from very close to the loop power supply (26 V) to a safe value compatible with the maximum allowable input voltage of the A/D converter, which is 5 V. Notice that the detector can exhibit four states, short- or open-circuit, normal and alarm condition on a single line.

The MC68HCllA4 has on board a programmable serial communications channel (SCI) with its own programmable baud rate generator that can support transmission speeds from as low as 75 to 38400 bauds, and variable word configuration. In cases where two serial communications channels are required, another serial communication channel could be synthesised by software using the onboard timer and the interrupt request line to detect the start bit in case of receiving. Of course, the processor has full control of the timing in case of transmitting. This arrangement dictates that the central station has also at least two serial communications channels (UART), a very easy task in the light of today’s LSI sophistication.

4.1.I Supervised high current output devices: Provision is made for three high current output devices. Each output is driven by a Darlington power transistor which is controlled by a bit from output port B. Terminal A of the output device is connected to the microcontroller A/D converter by an R/5R divider resistor network. When the A/D channel is read, it relays to the processor the state of the output device, a feature useful in protecting the Darlington transistor beside detecting output faults.

4.1.2Unsupervised inputs: There is provision for five unsupervised inputs, which could be the detectors’ local reset inputs. These inputs are connected to an input port via pull-up resistors.

4.1.3Unsupervised outputs: These outputs are low current output devices (buzzers and LEDs). These are of the open collector type which requires a low power driver transistor like the popular 2N2222. Also, like the unsupervised inputs, there is provision for five unsupervised outputs which can serve as local alarm indicators at the base of each detector. 1t * F

74C14

. -.

4.3 Communications protocol The communications link between the central station and remote field microcomputers (local stations) has to be very fast and reliable but also simple. It has to be noise tolerant. In normal transmission systems a penalty is assigned to the wrong choice of accepting a message and an algorithm is applied to minimise a loss function. In fire monitoring systems, because of the risk involved, there is no room for accepting a wrong message; the penalty assigned to the wrong choice is great, and the loss function must be zero. In this implementation, a class of error detection technique, the cyclic redundancy check or the check-sum method plus the usual hardware parity checking, has been employed. This two-way error detection scheme reduces to a bare minimum the possibility of erroneous message transmission between the local controllers and the central controller. Basically, to establish a reliable communications link between each of the local microcomputers and the central microcomputer there are two ways of connecting the field microcomputers with the central microcomputer. Fig. 2 illustrates the first way, called the open-line method, and Fig. 5 illustrates the second, called the closed loop, method. In our design the open-line method has been implemented. 5

Central station t o local station

Table 2 illustrates the message structure for this particular structure. Bit 1 of byte 1 is zero, indicating that the flow of data is from the central station to a remote 2N2222

4 . m

IT TRANSMITTER

Lrl

IN914

2M

2N2222

TO SERIAL RECEIVER

24kn

TWISTED PAIR

6

-

Fig. 6

Transmission line drives

I E E PROCEEDINGS, Vol. 135, P t . G , No. I , FEBRUARY 1988

Table 2: Central station message structure _

_

_

_

Bit 1

_

~

Bit 2

Bit 3

Bit 5

Bit 4

- - Messaae lenath - - - - - - - - - - -

Bit 6

Bit 7

Bit 8

Command code - - - - - - - - - -

- - Second command byte required only for output turn on - Check sum byte

station whose I D code is contained in the following seven bits (bits 2 to 8). The first three bits of the second byte contain the message length which in this case will be two except for turn output on command which will be three. The next five bits of the second byte are the command codes, yielding 32 available commands. However, only six commands are utilised for the time being. Except for turn output on command, the third byte will be the check sum byte. The fourth byte will be the check sum byte in case of turn output on command. If the message length is increased the structure of this message can be changed in such a way that bits following bit 1 are indicative of the length of the message. 5.1 Command code The central station sends six different commands to the local station, summarised as follows: 1 Reset and reinitialise 2 Go to sleep 3 Wake up 4 Send status 5 Cut detector power 6 Turn output on

covered, including : Detector in short circuit Detector in alarm condition Detector in normal condition Detector in open circuit Output in short circuit Output on output off Output in open circuit. 5.3 Message timing and response time From the previous discussion for the design under consideration and with reference to Tables 2 and 3, the longest message between the central station and the remote station is when status is requested, which is three words for the command and six for the response, a total of nine words. If a word is equal to 11 bits, then 99 bits are required. Assuming that the transmission speed is 9600 baud, then the message time is given by

Message time = (99 x 1000)/9600ms per remote = 10.3 ms per remote

Thus if we have say 128 remote stations, the central station will detect a fire in as little as 10 ms if the fire was reported by the first remote or as long as 1.32 s if the fire was reported by the last remote. 6

Main loop

Fig. 7 illustrates the software used with the local microcontroller (local station controller). After power-up, the initialisation phase starts by disabling all interrupts,

These commands are used to direct the local station’s operation for its optimum performance in terms of saving in power, sending and receiving information and detection of failure at a line or local station or any output. In this software architecture provisions are made to control various outputs through bit manipulation of the data byte.

START

I

READ DETECTORS STATUS STORE IN STATUS RAM

5.2 Local station to central station The local station transmits to the central station only when it is instructed to send status. Otherwise, it simply executes the command sent by the central station. Table 3 illustrates the message structure for this particular example.

Bit 2

Bit 3

- - Message length

--

Bit 4

Bit 5

Bit 6

Bit 7

t

I

Bit 8

READ UNSUPERVISED INPUTS STATUS STORE IN STATUS RAM

I

- - US1 1-US1 5 status - -

Det. 1 status

Det. 2 status

Det. 3 status

Det. 4 status

Det. 5 status

SO 1 status

SO 2 status

SO 3 status

- - not used - -

SEE FIG.8

READ SUPERVISED INPUTS STATUS STORE IN STATUS RAM

Table 3: Remote station message structure Bit 1

f INITIALIZE PORTS, CLEAR RAM CLEAR COUNTERS. ESTABLISH SP

I

- - US1 1-US1 5 status - Check sum byte

SO = Supervised output U S 0 = Unsupervised output US1 = Unsupervised input

OUTPUTUNSUPERVISEDOUTPUTS COMMAND

c

I

E SLEEP MODE FLAG SET?

Provisions are made in software for the direction, length and address of the message, where the first bit indicates the direction of transmission followed by the length of the message indicator and the message, plus any error correcting code if applicable. The data bits have the encoded status of the local stations. Most situations are I E E PROCEEDINGS, Vol. 135, Pt. G , No. 1, FEBRUARY 1988

WAIT

Fig. 7

Main loopJlow chart 7

>

I 1

8

IEE PROCEEDINGS, Vol. 135, Pt. G , N o . I , F E B R U A R Y 1988

loading the stack pointer, initialising the 1 / 0 ports, clearing the RAM thus clearing all software counters and setting all flags to the normal condition. After the initialisation phase, the local controller executes a continuous loop. First, it reads the status of the detectors and stores it in a special RAM area that is reserved for device status. It then reads the output devices status and saves the data in RAM, then reads the status of the unsupervised inputs and again saves the reading in RAM. After reading the status of all the devices connected to the microcontroller, it reads the supervised output command flags from RAM and turns the supervised outputs on or off and then repeats the same procedure for the unsupervised outputs. The processor then checks the wake/sleep mode flag and if it was set to the wake-up mode it executes the same loop again. If, however, the sleep mode flag was set, the processor executes a ‘wait’ instruction which in effect places the processor in a low power mode until it is interrupted by the UART.

that the buffer is not empty) and decrements the buffer counter. If the local microcontroller was interrupted by the receiver, then it reads the receiver shift register byte, and stores it in the receiver buffer and increments the message counter. If the received byte was the second byte in the message, the message length will be extracted from the first three bits. If the received byte was the last byte, then the message counter is cleared to get ready for the next message and the message is checked for errors. If the message was error-free, then the command will be executed, ortherwise the message is ignored. 6.1.2 Command execution: Six commands are executed depending on flag bits within the message. These are given in Section 5.1. 6.1.3 Timer overflow interrupt handler: The timer over-

flow interrupt handler turns on the pass transistor, thus supplying power to the detectors and ending the onesecond interruption of detector power. It also disables the timer overflow interrupt.

6.1 Interrupt handler routines 6.1.1 UART interrupt: The UART is the main source of

interrupt. The flow chart (Fig. 8) illustrates how the local microcontroller reacts upon being interrupted by the UART. If it was interrupted by the transmitter, the local microcontroller loads the transmitter holding register of the UART with a byte from the transmit buffer (provided

7

Example for a fire detection system

The central station resides in a control room and should have the means for proper signal identification; it provides subsystems for display monitoring and recording, circuit fault indication and power supply monitoring. Furthermore, all fire suppression systems should also be

CENTRAL STATION

I

I

MIMIC BOARD

MIMIC BOARD SAMEAS

1

2 1

LOOP LINE

CRTTERMINAL

I

I

MIMIC BOARD

S A M E AS

REMOVE STATION

I

4 1

DETECTORS

REMOTE STATION

w I E E PROCEEDINGS, Vol. 135, Pt. G , N O . 1, FEBRUARY 1988

Fig. 9

System architecture

9

monitored by the central station, such as waterflow alarm storage tank level, etc. Fig. 9 shows a system which consists of the central microcomputer station, several mimic boards distributed along key locations in the building (such as the superintendant’s ofice), a CRT terminal for system fault diagnostics if required and finally the local stations connected to the central station via a single cable. The central microcomputer could be the 280 microprocessor running at 4 MHz with 16 Kbytes of EPROMS and 8 Kbytes of RAM. The memory requirement varies from one system to another depending on the system specifications and complexity. In this example, the circuit has three SIOs (Serial Input/Output) for a total of six serial communications channels. The first four serial communications channels are dedicated to the four mimic boards, while the fifth communicates with the diagnostics terminal and the sixth is reserved for communication with the local stations. Note that, owing to the distribution of decisions, there is a low demand on the central microprocessor, hence this can be replaced with a microcontroller as well. The LEDs on the mimic board point to the location of a fire or a fault. If a detector goes into an alarm condition which is detected by the central station, the central station sends a command to the mimic boards to flash a common fire LED, turn the buzzer on and flash the LED corresponding to that detector. This action continues until the fire is acknowledged via the accept switch of the mimic board indicating that the fire has been detected by the people in charge. In this example, the mimic board also has its own 280 microprocessor since it may be located far away from the central station. 8

Conclusion

This paper describes the development of a large scale fire detection and alarm system using multi-single chip microcomputers. The architecture used is a two-level hierarchy of decision making. This architecture is made possible by the new CMOS microcontrollers which represent a high packing density at a low power consumption yet are powerful in data processing and thus in decision making. Each local station could make an autonomous decision if the higher level of hierarchy allows it to do so. It has been tried to keep the system design in general format so it can be adapted to varying situations. A prototype of the described system has been built and tested [lo]. The control part of the central station is implemented with a development card based on MC 68000 microprocessor (MEX 68KECB, by Motorola), which has a built-in monitor called Tutor. The application programs were developed using the features provided by this monitor. The local stations’ controllers were designed using the MC 68705R3, single-chip microcontroller. 9

References

1 ‘Fire protection guidelines for nuclear power plants’, US NRC Regulatory Guide 1.120

10

2 BAGCHI, C.N.: ‘A multi-level distributed microprocessor system for a nuclear power plant fire protection system controls, monitoring, and communication’, IEEE Trans., 1982, PAS-101,pp. 3 0 3 6 3043 3 PUCILL, P.M.: ‘Fire hazard protection, detection and monitoring systems’, Sea. Con, 2, Proceedings of Symposium on ADV in offshore and terminal measurement and control systems, Brighton, England, March 1979, pp. 353-363 4 HEAVISID, L.: ‘Offshore fire and explosion detection and fixed fire’. Offshore Technological Conference, 12th Annual Proceedings, 4, Houston, Texas, May 1980, pp. 509-522 5 CELLENTANI, E.N., and HUMPHREY, W.Y.: ‘Coordinated detection/communication approach to fire protection’, Specif: Eng., 1982,47, pp. 58-62 6 ‘Motorola Microprocessors Data Manual’ (Motorola Semiconductor Products, Austin, Texas, USA) 7 Electronic Industries Association : ‘Interface between data terminal equipment and data communication equipment employing serial binary data interchange’ (EIA Standard RS-232, Washington, DC, 1969) 8 MESAROVIC, M.D., MACKO, D., TAKAHARA, Y.: ‘Theory of hierarchical multilevel systems’ (Academic Press, 1970) 9 KASSEM, M.: ‘Fire alarm systems’, MSc. thesis, Dept. of Elec. & Comp. Eng., Concordia University, Montreal, Canada, 1985 10 LIE, P., and KOTAMARTI, U.: ‘The design of a fire alarm system using microprocessors’, C481 Project, Dept. of Elec. and Comp. Eng., Concordia University, Montreal, Canada, 1986

Asim J. AI-Khalili received his MSc and PhD in Systems Engineering from Strathclyde University (Glasgow, Scotland) in 1974. He previously worked with GEC-Elliott Automation and Parson’s Brown International as a Systems Engineer. Currently, he is an Associate Professor of Computer Engineering at Concordia University, Montreal, Canada. His interests include microcomputer systems architecture and applications, VLSI architecture, VLSI design automation.

Dhamin M. AI-Khalili obtained his BSc degree in electrical engineering in 1966, and the MSc and PhD degrees in digital electronics from the Unversity of Manchester, U.K., in 1970 and 1972, respectively. He joined the Ontario Centre for Microelectronics (OCM) in September 1982 as a Senior Engineer. He was involved primarily in carrying out and managing projects on the design of CMOS semicustom ICs. Also, he was involved in coordinating the technical activities of the Canadian Semiconductor Design Association as a Research Director and carrying out projects on cell library database generation. Before joining OCM he spent two years with Winnipeg Macroelectronics Centre as a Senior Engineer and a Project Leader in the area of microprocessor applications.He was formerly Director of Electronic Research and Development Centre in Baghdad; and Associate Professor and Head of the Department of Electrical Engineering, University of Technology, Iraq. He carried out research projects in both digital and analog applications. He is presently with the Department of Electrical Engineering at the Royal Military College, Kingston, Ont., Canada.

IEE PROCEEDINGS, Vol. 135, P t . G , No. I , FEBRUARY 1988