FUSE Project Burglary Detection: Digital Signal Processor (DSP) improves accuracy and reliability of intruder detection. Demonstrator Document

Transcription

1 FUSE Project Burglary Detection: Digital Signal Processor (DSP) improves accuracy and reliability of intruder detection.

2 Abstract The company Listener Sicherheitssysteme GmbH was founded in 1996 and is a smallsized company with 5 employees. The main business field of the company is the development, installation, distribution and maintenance of Burglary Detection Systems and other security systems for private as well as commercial applications. The company distributes own products and systems of other manufacturers as well. Main product is the LISTENER burglary detection system family consisting of a basic device that can be combined with additional units: The implementation of additional sensors (conventional contact elements, movement detection) is possible if necessary or requested by the customer. The existing device detects burglaries on the basis of analog processing of characteristic signals in the infrasonic frequency range. It can be used for flats or small detached houses. Objective of the Application Experiment was the improvement of the competitive situation of the company, substituting the old, analogue based solution by digital data processing methods. For this purpose, the DSP technology was introduced to our specialists within the Application Experiment. Using this technology and the corresponding software developed in this project, restrictions in the analysis of the very complex input signals that were existing up to now, were overcome. The in-depth analysis of burglary-typical signals and their differentiation from normal disturbances is now possible and guarantees a higher system reliability and the prevention of false alarms. Buildings, offices and flats up to an area of 400m² can be now supervised. Within the specification task, the DSP AD2181 from Analog Devices was selected, because this component met all our requirements, having a reasonable price at the same time. Listener s microelectronics experiences were restricted to overall knowledge in the design of analogue and digital systems using standard components and PCB design in throughhole technology before. Experiences in microcontroller or DSP technology were not existing, neither from the management nor the technical point of view. We are now able to plan and to manage such projects, to develop DSP based systems, to handle the corresponding development tools, and have extended our PCB design knowledge towards SMT technology. The project was started in March 1998 with a planned duration of 9 months. In the evaluation task of the prototype it became clear that more effort had to be spent than planned before. The parametrisation of the system making it reliable for all possible applications was more time-consuming than expected. Other unforeseen businesses caused an additional delay, resulting in a necessary total project extension of 3 months. The total cost of the experiment to industrialisation was 77.3 K of which 54 K was the FUSE investment and 20 K was for industrialisation. With the anticipated sales increase and the resulting profitability improvement a payback period of about one year and an ROI of 700% over 5 years of sales is expected.

3 Keywords DSP technology Burglary detection Infrasonic measurement Alarm systems Small company Signature D 1. Company name and address LISTENER Sicherheitssysteme GmbH Zschopauer Straße Chemnitz 2. Company size Listener Sicherheitssysteme was founded in 1996 and has currently a staff of 5 employees: Personnel description: Function Manager* Management assistant Distribution Development engineer* Manufacturing Qualification university diploma, field of study machine building, specialised in medical equipment university diploma, process control engineering and manufacturing control Business employee University diploma, engineer for electrical engineering Skilled worker, electrical engineering * involved in Application Experiment If necessary, additional employees are working in the distribution and installation of the burglary detection systems. The manager and the development engineer were mainly involved in the Application Experiment. The company is fully independent and has a turnover of about per year.

4 4 3. Company business description The company Listener Sicherheitssysteme GmbH was founded in May 1996 having the name BIZEQ at this time. We also started the Application Experiment as BIZEQ GmbH. In 1998, the company was renamed in order to make the main business field of our company more obvious. Listener s main business fields are the development, distribution, installation and maintenance of burglary detection systems and other security systems for private as well as commercial applications. The company distributes own products and systems from other manufacturers as well. These third party products belong to the group of conventional acoustic or optical surveillance devices, e.g. for security services. Another business field is the video equipment sector for different applications, mainly in the security sector again. As already stated, the company's main product is the LISTENER burglary detection system developed and owned by the company. The share of the total turnover is about 80%. Our company with 5 full-time employees has not the capability to manufacture the systems in-house. The production of the mechanical parts (e.g. cases) and the PCBs is outsourced to regional partners. At the company s site, the LISTENER systems are assembled and tested. The basic systems can be completed with additional acoustic and data transmission modules according to the needs of the single customer. The distribution and sales activities are two-fold: On the one side, we install and offer our products in the regional area ourselves. The maintenance service is also guaranteed by the company. This enables us to react very fast to the requirements of the customers, creates at the same time an atmosphere of confidence and results in a kind of thirdparty advertisement - free of charge. At the same time, we become familiar with the technical requirements and actual market trends. On the other side, we are not able to cover the whole German and international market. That s why we signed contracts with several distributors located in nearly each region of Germany, but also in Switzerland, Italy and Austria. They are trained in an initial course in order to be able to explain, to install and to maintain our products. Documents and updates are supplied continously. This network was extended especially in the last two years and will be also the basis for the projected sales figures of our new product. 4. Company markets and competitive position at the start of the AE As described above, the main business fields of Listener Sicherheitssysteme are development of burglary detection systems and its distribution, installation and maintenance. Hence, the company's products belong to the market segment Electrical equipment (Prodcom Code 31), especially burglar and fire alarms (3162).

5 The market for burglary detection systems is divided into two sectors: public authorities (banks, administration, prisons) and surveillance of big industrial complexes on the one side and the private/retail shop/sme sector on the other. 5 Regarding the first sector, the market is nearly saturated, systems can be sold only for new buildings or when substituting an old installation. However, this is not the business field of Listener. Typical customers in the private sector (in which our products are mainly sold) are owners of detached houses. In the commercial oriented retail shop or SME sector, sales figures up to now were limited because of the technical product features of the old system. Potential customers are here owners of shops or workshops. This market is extremely growing. The percentage of the total turnover of the company in the security sector is in the range of 90%. This clearly demonstrates the importance of the AE for Listener s business. The market for our second main business, the video equipment sector for different applications in the security sector again is also very promising. However, these application-oriented activities are limited to our region, because they require a lot of services to be provided at the customer's site. We estimate here a potential regional market size of about 2 M per year. With about , LISTENER has a market share of 5% in this field. Competitors on the market are mainly big companies, like Siemens, Bosch or ABB. Their products are normally in the high price range (between and ) and it requires a lot of effort to install the systems. This prevented in the past a common use of such systems. From the technology point of view, most of the systems were based on conventional discrete analog and digital components, some of the systems contained a microcontroller. The existing Listener V3 with its volumetric working principle and the related measuring principles has clear cost advantages. On average, the Listener V3 was sold for 750. Installation and service effort is minimal, the customer can install and adjust the unit without problems in a few minutes, if he wishes. This opened us a market segment at least in the regional area, but also in Germany. The turnover reached about per month. With the existing product Listener V3 our company was able to enter into a very fast growing market segment. The criminal rate is growing, the danger of a possible burglary is increasing more and more. On the other side, only about 4% of potential customers for burglary detection systems in the retail shop/sme sector, but also private area are equipped with an alarm system up to now. That's why the demand from both sectors is growing in the last years continously. Actually, we estimate for Germany a total market size of about 25 Mio. per year with increasing tendency. Listener Sicherheitssysteme had a market share of about 10% in the regional market and about 3% in Germany. Products were sold to other European countries, e.g. to the Ukraine and Czech republic, Switzerland and Austria. However, this market share is negligible. On the other side, the Listener V3 could be sold only to owners of flats or small detached houses. The future market growth in the commercial sector would have been without influence on our sales figures. An additional tendency is dangerous for our market position and might result in a decline of our market share: The big companies competing with us offer more and more also low price solutions and penetrate the market very fast.

6 6 The existing product Listener V3 with its limited application area, a restricted functionality (only one sensor, limited parametrisation capabilities) and Not-100%-security against false alarms resulting from the analogue measuring and signal processing principle is not able to compete with the market leaders in the future. We have to expect the following decline in the market without product improvements: Forecast of turnover without product improvements (1997=100%) Turnover (in %) Year Turnover in% Therefore, the objectives for the new product Listener V5 are as follows: Substitution of the analogue signal processing principle by a digital one Increased functionality and safety against false alarms Extension of the application area to big detached houses, offices and commercial buildings by introducing an optional multisensor system Improvement of communication capabilities and modularity The features mentioned above will enable us to extend our market share essentially in the private sector and to have additional customers in the commercial area for the first time. Costs are not the main driver, they will be nearly the same as before. The basis for the calculation of the ROI of the Application Experiment will be not the cost saving, but the additional units sold. 5. Product to be improved and its industrial sectors Listener Sicherheitstech GmbH is on the market with its Burglary Detection System Listener V3 since The device uses volumetric principles for the detection of burglaries in enclosed spaces like offices, flats or appartments. A continous background noise level analysis is implemented to prevent false alarm functions. Furthermore, an interface for the external connection of additional signals (e.g. door contacts) is available. An implemented siren delivers in an emergency case an acoustic signal of 120 dba. Listener V3 measures and analyses continously pressure changes which are related to changes of material consistencies and tensions which are typical when forcing one s way into flats or houses. If there is a door broken open or a window is smashed or taken off its hinges, in any case Listener V3 detects the infrasonic signals which are related to such an event and triggers off an alarm signal immediately. The device is able to distinguish

7 7 between the normal infrasonic signals resulting from continous movements inside the area to be monitored and spontaneous events coming from unauthorized access. The detection of the burglary already in an early phase is very important. In this way, the burglar is disturbed by the alarm signal already before entering the house or flat. For this purpose, Listener V3 works very precise and reliable and can be installed mainsindependent. In combination with a connrected external alarm system Listener V3 offers highest security up to 250m² enclosed space. Technical parameters Type: Volumetric sensor device with two outputs for external units, external ENABLE-Signal Package: Aluminium cover, steel bottom (powder coated) Storage temperature: C Operating range: C Performance/Battery: Life & Service Cycle for battery 15 months 6 batteries (alcaline manganese, industrial type) Performance/siren: 120 dba Adaptation Period: 2 minutes Reset: automatically Disable time: 30 sec Sensitivity: adjustable The block diagram of the product is shown in Figure 2: Infrasonic analog Discrimina- timer, dis- Signal sensor filter nator crete data adaptation unit processing logic Output alarm signal Background- Noise Analysis Interface to external devices Fig. 2: Block diagram Listener V3 Advantages of the existing solution are the low component costs and the minimum measuring and test effort. The system is only hardware based, no software is implemented. Up to now, only standard components were used. The applied PCB technology (one layer, only through-hole technology) does not match state-of-the art technologies and is oversized. The main disadvantages of the existing solution are: the limited ability of the system to distinguish between burglary-typical signals and normal infrasonic signals resulting from continous movements inside the area or other noise, the high influence of the signal amplitude in the signal analysis

8 This causes a dependence of the evaluation result on the distance of the infrasonic event from the device. unsufficient self-test capabilities limited possibilities to interface additional sensors or standard surveillance components as input or additional alarm equipment as output 8 With the existing technical solution the extension of the application range to surveillance systems for bigger detached houses, industrial or commercial buildings, shops, workshops etc. is not possible. The implemented analogue circuitry cannot meet the requirements in this field and is not reliable enough. Therefore, it is necessary to use methods and algorithms of digital signal processing and analysis. The old solution (a 1-layer PCB in conventional Through-hole-technology) is shown in figure 3. Fig. 3: The old 1-layer PCB in through-hole technology 6. Description of the technical product improvements The new product Listener V5 is characterised by improved technical features and additional implemented functionality. The substitution of the old analogue filter elements in discrete technologies by digital signal processing methods made it possible to improve the following product parameters essentially: More reliable detection of signal curves which are typical for burglaries Extension of the room or building size that can be surveilled by one unit. The device is now also applicable for bigger detached houses, offices etc. Implementation of additional sensor elements is now possible. Detection of dangerous events in the complete enclosed space (e.g. house) Elimination of false alarms Interface compatibility to all state-of-the-art burglary detection systems resulting in a more flexible application of Listener V5 The following alarm cases can be processed now: - Burglary detection (4 lines) - Fire/Smoke (1 line) - Emergency call (1 line) - Sabotage (1 line) - Additional two inputs lines are reserved for external locks, keypad signals etc. Each line can be connected with up to 20 devices (total power consumption

9 9 must be less than 1 A) Ιntegration of sabotage protection mechanisms (protection against unauthorized disabling of the device) Further improvements are: Self-test capability Complex logic block for the processing of additional external control signals Power back up with accumulator charge control (protection against power failure) Extended interface for the connection of additional external alarm devices Additional operating mode External surveillance Application specific parameter setting via PC interface The block structure of the new system is illustrated in Figure 3, the pictures 4 and 5 show the new PCB and the system itself. Accumulator Voltage Control Unit EEPROM Configuration Data Digital Signal V24 Processor Serial Interface AD 2181 I Processing Interface for of external alarm activation alarm network ADU-Block signals Analog signal preprocessing Analog alarm sensor signal Siren Flash Telephone Display Microphone Fig. 3: Block structure of Listener V5 The important step towards the improvement of the system is the change from analog to digital signal processing methods. In the rooms/buildings which have to be monitored a sensor (microphone, barometric pressure sensor) is installed. The resulting signal curve coming from the ADU block is analyzed in the DSP. The differentiation between alarm relevant and other noise signals is now based on the processing of digital algorithms that are nearly independent on the amplitude of the signal.

10 10 In order to increase the reliability of the system a detailed frequency analysis using FFTalgorithms which detects typical alarm curves is implemented. Compared with the old solution, the monitored infrasonic frequency range was extended. The following parameters in the amplitude-response are relevant for the alarm case: rise of pulses, plateau and decay characteristics of the curve, and the harmonic content of the signal. That s why an analysis of the time-dependent behaviour of the digitalized alarm sensor signals is executed. This must be done under real-time conditions. The result is compared continously with the actual level of background noise in order to prevent false alarms. In case of an alarm situation, the DSP controls the interfacing to external alarm systems. The optional implemented interface to police or private surveillance organisations increases the safety level for the customer essentially. Figure 4 shows the new DSP board, figure 5 the new LISTENER as it is installed in the buildings. Figure 4: The new DSP board

11 11 Figure 5: The result of the Application Experiment: Listener V5

12 12 The following table compares the main technical features of the old and the new product: Manufacturing Costs per Listener Old Listener V3 New Listener V5 ~ 400 DM ~ 350 DM PCB one layer PCB in THT Two-layer PCB in mixed SMD/THT Computing performance Not existing High, application of fixed point DSP AD2181 Complex math. Algorithms not possible to implement Use of filter and data processing Algorithms, FFT Data memory Not existing 160 k Analog to digital Analogue processing on-board ADC conversion Testability poor, offline hardware test if necessary excellent through implementation of self- test capabilities Application Smaller detached houses, mainly private sector All kinds of detached houses, but also commercial sector Flexibility very restricted, fixed hardware structure High flexibility because of full parameter set Table 1: Technical features of old and improved product The realisation in mixed THT/SMD technology is an essential technology improvement. However, the main driver for increased sales figures will be the functional improvements and the increase of reliability resulting from the use of digital signal processing algorithms. 7. Choices and rationale for the selected technologies, tools and methodologies At the beginning of the project we had to select the best technology for its realisation. We had to consider for our decision the following conditions and requirements: Medium number of pieces (<3000 per year) Reasonable costs for software tools and hardware Fast prototyping Necessary flexibility to modify the basic system corresponding to the customers requirements and the structure of the building or office to be monitored (e.g. parameter setting according to the geometry and size) Ability to handle similar tasks without outsourcing in the future Character of next planned projects (in order to replicate the know-how) Real-time processing of event ability to process difficult mathematical algorithms (e.g. FFT)

13 13 For the realization of the Application Experiment different basic principles were possible: software based solution with microcontrollers or DSPs implementation of the digital signal analysis algorithms and control logic in a FPGA or ASIC A further development of the discrete solution with conventional analogue and digital components was not possible. The following matrix visualizes our premises in relation to possible technology choices: Medium number of pieces Digital ASIC FPGA DSP Microcontroller Costs f. tools Fast prototyping Flexibility Software modules -- (µc core possible) No outsourcing High speed requirements and difficult algorithms Table 2: Technology selection criteria The implementation of timing and frequency analysis algorithms into a customer specific hardware (ASIC) would result in a very high complexity of logic and memory structures. The advantage of such a solution are very fast reponse times, but due to the high number of gate equivalents and the pin count of the chip the solution is too expensive, especially when regarding the implementation of time and frequency based analysis and the medium number of pieces needed per year. The necessary software algorithms to be processed eliminated a FPGA solution. The analysis tasks to be done are typical software problems which can be optimal implemented in microcontroller or DSP structures, respectively. Additional arguments are the comfortable possibilities to adapt the system via parameters in the installation phase. Taking into consideration the real-time analysis of the sensor element signals the FFT analysis in the frequency range the control tasks for the whole system in general and especially the interfacing to the external alarm network it was decided to use a DSP for the Application Experiment.

14 14 The DSP has to process the FFT. After that, the energy densities of the frequencies are weighted and analysed. That means that in the Fourier domain of the time function the current data are compared with reference values in order to define alarm states. Furthermore, the DSP realises all logic functions of the burglary detection unit digital filtering for the analog signal processing digital chatter-proof sequencing control for external signalling systems signal modulation for internal signalling systems generation of control signals for optical display As defined in the submission, the TMS320C50A was our first choice. However it became clear in the specification task that Texas Instrument DSPs did not fit our functional requirements and the foreseen price range ideally. That s why DSPs of other manufacturers were considered, e.g. Motorola or Analog Devices. The final choice was between the TMS types and Analog Devices, especially the AD The advantages of the AD2181 are the following: Component price 40 DM for TMS plus additional 15 DM for external memory against 20 DM for AD 2181 (memory is already implemented) Better EMC behaviour of Analogue Devices DSPs because of lower clock frequencies and internal memory The internal memory organisation in form of a ring storage is ideal for real-time applications. Additionally, the selected DSP type is: (more than) sufficiently powerful for the planned applications well supported by tools and manufacturer stable available from stock In order to guarantee the error-free development of programs and to be able to test the hardware under realistic conditions, a real-time emulator must be used. This allows a program test with simulated worst-case-conditions. Depending on this decision the necessary PC based tools (debugger, Assembler/Linker, C compiler, emulator) were selected and rented from the subcontractor. For a next development, the C-compiler and debugger software will be purchased by Listener Sicherheitssysteme, in order to allow replication. Possible supplier of the tools are distributors (such as SPOERLE in Germany) or Analog Devices itself. In detail, the ADSP-2100 family development system was used consisting of System builder (definition of archtitecture for the hardware system) Assembler/Linker Simulator PROM Splitter C-Compiler In addition, the EZ-ICE emulator for hardware based debugging was used.

15 15 As discussed above, in the existing product conventional PCBs in through-holetechnology were applied. The PCBs were designed and manufactured by external partners. A second, more peripheral objective of the Application Experiment was therefore to become able to design our PCBs in the future ourselves using a PC based tool. This will allow a higher flexibility and reduces design costs. In the Application Experiment the system Eagle with integrated autorouting system was used. The reasons were the following: reasonable PC version with high functionality 32 bit and coprocessor support error check in real-time already in the design phase flexible layout parameters for EMC compliance convenient generation of error-free manufacturing data (including different checking algorithms) generation of different layout versions for one design permits several assembly and technology versions of one design (--> higher flexibility of one PCB, reduction of costs) Furthermore, a detailed user documentation and support in German is part of the Eagle system. Another aspect is that the system is introduced to the market. It is often used in the development of systems similar to our project and is also applied by our partners. The output data format is accepted by all PCB manufacturers in the region. The programs and boards were tested in all phases of the project. After each design milestone, an internal review with the subcontractor was made. This resulted in the awareness that each new task could be started on a checked intermediate result. This helped to avoid additional iterations. The field test of the whole device was of special interest. In order to be sure that the device can be parametrised for all applications we made a field test under real conditions. For this case, rooms in demolition houses were rented and burglaries simulated. Some parameter ranges had to be modified without changing the algorithms significantly. Nevertheless, a higher effort than expected before was spent in the evaluation task. 8. Expertise and experience in microelectronics of the company and the staff allocated to the project Listener Sicherheitssysteme is an experienced developer and distributor of burglary detection systems. The company has know-how in application-specific system and product designs in this field. The described Listener V3 demonstrates very clear the experiences in the application of volumetric principles. There is no other competitor on the market offering devices with the same technical principle. We were able to design such systems on system level and to develop a corresponding realisation on paper using discrete analogue and digital components. The layout design of the PCBs was outsourced, the manufacturing and the assembly of the PCBs as well. Before the Application Experiment, the staff of the company and especially the two involved specialists had basic knowledge in the field of programming and microelectronics technologies (resulting from their studies). The development engineer had additional know-how in the design of analogue and digital systems using discrete standard

16 components. 16 We had no experiences in the realisation of microcontroller or DSP based project developments, neither from the management nor from the technical point of view. Know-how about the current DSP market situation was not existing, such as available types, technical features and conditions for applications, prices integration of DSPs into the product software technology (program development in assembler/c, test and debugging strategies etc) Because of this lacking knowledge we were also not able to manage such a project. So it was not possible to setup a realistic workplan (no estimation of real effort was possible), select appropriate DSP types and tools (missing overview and no selection criteria available), evaluate our own capabilities (possibility to realise the project together with a subcontractor), consider all main criteria for a later serial production In general, the former projects of the FU never exceeded a duration of 3-6 months. Going beyond the scope of the DSP application, there was no experience in the planning and structuring of a longterm project like the Application Experiment, and it was unknown how to manage this kind of task. Obviously a certain amount of project management training was required, especially - but not only - for DSP projects focussed at project planning, workplan, selection of tools, interfaces to a subcontractor and project control/management. 9. Workplan and rationale In the preparation period of the submission a workplan had been discussed with the TTN/subcontractor that was the basis for the work executed in the Application Experiment. Because of the personal engagement of all people involved in the Application Experiment and the good assistance by the subcontractor who reacted very fast to all of our sometimes unforeseen - needs there were no essential deviations between planning and realisation of the project except the evaluation and test activities that were more time and effort consuming than expected. The reason was, that one of our objectives was to make the device suitable for all possible applications, starting from conventional detached houses up to offices and manufacturing buildings with a difficult geometry. This resulted in very complex field tests. In order have realistic conditions, we rented flats and halls in empty houses or factories that were foreseen to be pulled down. These buildings were not always available in time. That caused a delay of the schedule. Additionally, the tests resulted in some necessary changes of the system parameter ranges and slightly software modifications to make the system reliable for all geometries. One additional aspect should be also mentioned: Even in case of best planning of a project a small company has not enough reserves to compensate delays resulting from unforeseen activities in the daily business. In our case, the Christmas business was very intensive. That s why it was not possible for us to spend the effort in December as planned

17 17 before. One of the lessons learned is that additional time slots should be implemented as a form of contingency planning in order to be able to compensate such effects. The initial workplan and schedule which was defined in the beginning of the Application Experiment was as follows.

18 18 Month March 1998 Subject Training activities: -Technical management: Selection of technologies, definition of workplans, selection of technologies and subcontactors, special management issues (monthly reporting modalities, cost statements etc.) -Programming and application of DSPs, development tools Start of specification (selection of DSP type, resulting system configuration etc.) April-May 1998 System and block specification, I/O interfaces, definition of software parameters, structure of software modules (e.g. FFT) June-October Design of hardware system LISTENER V5, development of 1998 prototype board (including introduction to PCB design tool), software programming, prototype manufacturing, start of test and evaluation November 1998 procedures Continuation of test and evaluation activities Table 3: Project schedule (as planned) The original workplan is shown below. The dotted (blue) lines reflect the real schedule as realised in the project. The effort in the Technical Management was higher than expected because of additional efforts in reporting (preparation demonstrator documents), negotiations with the TTN to modify the workplan etc., and additional organisational work to find and rent the rooms in old houses and factories. The effort in the specification task was a little bit higher, because we wanted to use the best suitable DSP for our project. We had to inform about different possible DSP types and their advantages/disadvantages. This market survey and the assistance given by the subcontractor enabled us to select the best DSP type for our project. This was an additional work going beyond the scope of only specifying the system on the basis of a already predefined component, and a kind of additional, technology based management training. Besides other internal businesses the extension of the project for three months was mainly caused by the additional test and evaluation effort mentioned above. The software modifications in the test and evaluation task were already a kind of internal replication of the know-how. They were made nearly without assistance from the subcontractor and have verified the successful know-how transfer.

19 19 Task1: Technical management M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 Final Report Dissemination Task 2: Training Techn. Management DSP PCB Task 3: Specification System specification, selection of DSP Block and system structure, hardware Software specification, signal analysis Task 4: Design Hardware development Board design & manufacturing Software modules System Task 5: Evaluation DSP Test PCB Prototype Test Product test

20 The work in the different tasks was organised as follows: 20 Task Listener Sicherheitssysteme Subcontractor Management Project related management Support in technologyoriented decisions Training Participation in training Training Technical Management, DSP technology and tools, introduction to PCB design Specification Design Evaluation & Test Table 4: Division of work Function-related tasks Hardware structure of the system Software specification Hardware development of Listener V5 PCB design Specification Review Hardware design review Co-operation in technology related tasks, continous support Design assistance as necessary, application of Learning-by-doing methodology with decreasing level of assistance Software development and Introduction to programming, assistance as test necessary Software development review Prototype test and software modification in form of field tests under real conditions, parameter modification PCB manufacturing Assistance As can be seen in the table, the First User did all the specification, design and software programming as well as test work and was assisted by the subcontractor as necessary. The selected know-how transfer methodology was adapted to the permanent increasing knowledge level of Listener during the Application Experiment. At the beginning, the cooperation was close and direct (First User and subcontractor worked together). Later on, the subcontractor assistance was mainly focussed to answering questions of the First User and to review the intermediate results.

21 The following effort was spent: 21 Task First User Subcontractor Totals Role Effort Effort Cost Cost Role Cost Cost plan real plan (k ) real (k ) plan (k ) real (k ) Plan Management Training Specification Design Responsible Participant Responsible Real ,75 5,64 Assistance 0 0 3,75 5, ,0 3,22 Responsible 5,4 5,4 8,4 8, ,75 7,58 Assistance 2,2 2,2 8,95 9, ,5 10,96 Assistance, layout generation 9,4 9,4 19,9 20, ,5 12,1 Assistance 0,8 0,8 8,3 12,9 Responsible Evaluation Responsible Totals ,5 39,5 17,8 17,8 49,3 57,3 Table 5: Planned and real effort of the project In addition, about were spent for board prototyping and travel. Because the subcontractor is also in Chemnitz, travel costs were only a minimum. 10. Subcontractor information Listener Sicherheitssysteme GmbH had no experiences in microcontroller and DSP technologies and their application in product developments, neither from the management, nor from the technical point of view. Our intention in the Application Experiment was therefore to learn how to manage such a project and to become familiar with DSP programming methodologies, DSP based hardware design and its product-oriented application. Furthermore, we were interested to gain experience in test methodologies for DSPs. That s why we were looking for a subcontractor with the following competencies: experiences in training activities for DSP project management, tools and design/programming know-how in Learning-by-doing-methodologies and know-how transfer for this kind of work longtime experiences in cooperation with companies which are at the same technology level as Listener was know-how in different DSP types (independent choice of best suitable type), assembler and C knowledge application-oriented DSP know-how, references in DSP based product developments GEMAC Chemnitz was chosen as subcontractor. The company fulfilled the criteria mentioned above and has all necessary design tools available. The company is a SME with 55 employees, and has specialists which are experienced in all kinds of application-oriented design of customer-specific electronics in form of PCBs, hybrid circuits, MCMs, FPGAs, digital, analog or mixed signal ASICs and microsystems for more than 15 years.

22 22 From the beginning, one main business field of GEMAC was the programming and application of a wide range of microcontroller and DSP types from different manufacturers (TI, Analog Devices, Zilog, Siemens and others). This was very advantegeous for us. The training activities and the subsequent design assistance were therefore very effective. The subcontractor was also very flexible in his work. Depending on the actual requirements, occurring problems and the current project situation, a very dynamic style of cooperation was possible that helped us to keep the project mainly in time. The responsibilities and the division of work changed over the project. In the first training activities, we had to learn and GEMAC was responsible for the know-how transfer. Starting with the Learning-by-Doing" steps, the situation changed. The work was now characterised by a kind of decreasing assistance (which was given, whenever needed). The contract signed between the subcontractor and Listener Sicherheitssysteme covered all issues of the know-how transfer and all details related to the DSP design and the PCB prototype manufacturing as well. Especially, the following items were defined: Details and duration of the training courses (Technical Management, DSP overview, programming and application, PCB design) Details of design assistance (Learning-by-doing), duties and responsibilities according to the table given above Board manufacturing Project schedule and costs, DSP kit licences Listener Sicherheitssysteme owns the developed solution, the subcontractor has no commercial or scientific rights to use the technical results of the Application Experiment. Penalty clauses going beyond the scope of the common terms of business were not included in the contract. One additional reason for the selection of the subcontractor was that GEMAC as a regional partner of Listener Sicherheitssysteme will be able to give direct assistance in any future projects. The company is foreseen as manufacturer for the serial product Listener V5. This will be very advantageous not only because of all the technical competencies. As a result of the Application Experiment the subcontractor is already familiar with the detailed product solution. Any kind of information loss is therefore impossible. The detailed division of work and the spent effort was already described in the last section. It was a real cooperation between the partners. 11. Barriers perceived by the company in the first use of the AE technology Listener Sicherheitssysteme is a typical SME in East Germany which has similar problems as many others to improve its competitive situation and - related to that to innovate their products using new microelectronics technologies. The company - was founded as a new enterprise after the end of the GDR - had knowledge in the development of burglary detection systems and related equipment for different applications - had experiences in the design of digital or analogue systems on the basis of standard

23 components for this kind of application - subcontracted all PCB related development and manufacturing tasks 23 The manager and the development engineer had some basic knowledge in microelectronics. We recognised that the application of microcontroller or DSP technologies would be useful for the innovation of the company s products. Some new product features, e.g. the necessary flexibility, improved reliability, and the extension of the application could not be realised anymore with the old simple technologies available in the company at this time. Because of the lacking knowledge for the introduction of new technologies external assistance of an experienced subcontractor was absolutely necessary. Under real market conditions the next question raises: How to select the best one? Optimally, you have to find an independent subcontractor who is not fixed to a special DSP product (otherwise he will recommend you always the types he always uses, even if it is not the best component type for you). The same applies for the tool selection. Based on our experiences, Listener Sicherheitssysteme was already familiar with the management of design projects using conventional standard components. However, there was an information lack in the special issues of a DSP development (e.g. time schedules, price estimations, selection of tools and components). Of course we also had essential know-how barriers concerning best practice in DSPrelated hardware design methodologies, software development and product-oriented test strategies. Last not least it was not clear for us if we would really be able to do such a project ourselves starting from the specification up to the final product test (even if supported by a subcontractor). This was a kind of psychological barrier. Some of the barriers are discussed in the following more detailed: 1. Project planning and specification barrier Listener Sicherheitssysteme had no ideas about realistic cost estimations for such a project, about design risks, necessary design effort, component costs and availability of components. We had some ideas what should be implemented into such a system. But we did not know which capabilities the DSP technology really offers, i.e. if our objectives were realistic. 2. Project management barrier No reliable knowledge about the organisation and the schedule of a DSP project (including contingency planning) was available in our company. Furthermore we had no idea how to select the best subcontractor and how to realise best practice in know-how transfer. We had the problem to find a real independent subcontractor (not fixed to one DSP type). 3. Specification barrier Because we had no information about the capabilities of different DSP types it was not possible to write a specification for our new Listener product. This prevented contacts to a subcontractor before (you have to explain him what you want), an additional barrier why the project could not be started. Additionally, it was not possible to define a suitable division of work between Listener Sicherheitssysteme and a potential subcontractor. Missing experiences in the structuring of software-related DSP functions and remaining hardware functions as well as the specification of the hardware-software interface were additional barriers in this field.

24 24 4. Design barrier In the design we had a lot of very concrete technical know-how barriers. For instance, the following questions and problems arose: Which design methodology should be selected? How to make a design? How to write the software and how to test it? How to handle the tools? How to simulate the design? What is best practice for PCB design? 5. Test and evaluation barrier The company had insufficient knowledge how to test a DSP based prototype. Which kind of test sequences can be used? How to set parameter ranges? In which way a design iteration (in case of necessary modifications resulting from the test) must be done? There was another important barrier which is typical for small sized companies: The effort to introduce a new technology in the company could not be spent. The daily business, the design risk and the high costs prevented its introduction. Even if it would have been possible to go through a learning process starting from a training course, via first design steps in try and error mode up to a test of the developed solution, the effort, the risk and the costs were too high for us. Consequently, this resulted in a strong financial barrier that could not be overcome without external assistance as offered by FUSE. 12. Steps taken to overcome the barriers and arrive at an improved product Within the Application Experiment, the barriers mentioned above were successfully overcome: 1. Project planning barrier The process to overcome this barrier started already in the preparation phase of the submission. In several discussions with the TTN we got first knowledge about new innovative micro-electronis technologies, especially Technical and functional capabilities of microcontroller and especially DSP technology (including the corresponding selection criteria) DSP types, programming capabilities and development kits Realistic definition of workplan, costs, duration of the project Furthermore, a kind of feasibility study for our project ideas was made together with the TTN. The result convinced us that the DSP application was the right approach for our project, that the technology would be able to meet our technical and financial requirements and that it would be possible to learn the design methodology ourselves. The result was the submission, which was based on feasible technical and functional issues and realistic estimations concerning workplan, and costs. It included also a first calculation of the return-on-investment. The know-how transfer was continued in the Application Experiment itself. 2. Project management barrier In the preparation phase of the submission we discussed with the TTN the management issues of the project. This was continued during the Application Experiment. Problems

25 25 were immediately discussed during the monthly meetings. If necessary, the TTN supported the management. This was especially the case when modifying the workplan because of the unforeseen test and evaluation effort. Now we know how to manage a DSP development, what can happen within such a project and how to react in case of unforeseen problems. We are now familiar with best practice in this field. 3. Specification barrier In the training courses we got an overview of the most important DSP families available on the market and their technical main features. Based on this knowledge the final DSP type for our project was selected. Additionally, the training activities enabled us to structure the system in software modules and remaining hardware (e.g. for system interface, A/D conversion etc.). Very important for this know-how transfer was the used Learning-by- Doing methodology. We specified the system. If necessary the subcontractor assisted us in all technology related problems. A common review of the specification at the end of this task convinced us to have found the optimal specification for our project and to have overcome this barrier at the same time. 4. Design barrier Within the design task we got practical knowledge in software methodologies, programming, debugging and test methods as well as software implementation. On the hardware side, we learned to implement DSPs in the complete system and to design the peripheral hardware structure that is necessary to pre-process the input data. We also got an overview about PCB design methodology. Again the combination of training activities and immediate Learning-by-Doing-Steps was the most effective for us. 5. Test and evaluation barrier The way to overcome the test and evaluation barrier was two-fold. On the one side, we learned best practice in software debugging methods, on the other side we co-operated with the subcontractor in the PCB test in order to become familiar with test strategies in this area. The second part was the product test itself. We learned to adapt the system to real practical conditions and to do corresponding design iteration cycles. Because we did it more or less self-standing (only partially assisted, if necessary) it took us more time than expected. The positive result of this methodology is that we know now to be able to do this task ourselves. There is no barrier anymore. The initial financial barrier was overcome after the initial planning had shown what the expected ROI and payback figures were likely to be upon completion of the Application Experiment. 13. Knowledge and experience acquired Based on our knowledge in the functional design of burglary detection systems with standard components and conventional through-hole PCB technology it was our objective to become able to execute a successful DSP development including all necessary design steps selfstanding. The methodology to achieve this goal together with the subcontractor was already described. Based on the initial training courses and the own design and management activities assisted by the subcontractor we learned: