1. Electrical and Computer Systems Engineering Postgraduate Student Research Forum 2001 Design and Development of a Distributed Avionics System for use in Unmanned Aerial Vehicles Brad Phillips, Lindsay Kleeman, Andrew Price The Aircraft and its Work Conceived in 1992 as a solution for remote data gathering in cyclones and hurricanes, the Aerosonde is a small autonomous aircraft with a wingspan of around three metres. With an airspeed of 100km/h and an endurance of up to 40 hours the aircraft is perfect for missions classified as dull, dirty or dangerous. Such missions have ranged from cyclone studies and crop surveys to environmental monitoring in the Arctic. Because of the nature of its work, the Aerosonde is essentially considered to be an ‘expendable’ item. Regulatory agencies such as CASA and the FAA on the other hand rightly insist that the aircraft should not pose a hazard to other airspace users or to the public. The challenge is therefore to build an aircraft and avionics system that still addresses the expendability criteria while meeting the reliability expectations normally achieved by the addition of redundant systems. Current System Design The existing avionics is of a fairly classic design. A central processor (68332) provides the heart of the system. Interfaces to the rest of the world are through the various plug on modules such as the Signal Conditioner card and the System Motherboard. Apart from the recently introduced Engine Control Unit all sub modules are ‘unintelligent’. While a highly integrated system might appear from a design point of view to be an attractive solution, it poses significant problems for manufacturing and field maintenance. Without on- board intelligence, the testing of sub modules prior to final assembly is not possible without recourse to dedicated and often expensive test jigs. The complexity of a centrally focussed wiring harness makes detection of potential faults difficult in the field. A single processor means that all aircraft functionality is implemented in one software build. Consequently changes to one part of the system software can potentially have a global impact on performance. Management and testing of the software upgrades becomes problematic, especially as more people become involved in the project. In the current design software change control is further exacerbated by the continual incorporation of new payloads into the avionics system. A Distributed Approach The alternative to an integrated system is a distributed one. In such a system each node is capable of collecting and acting upon data independently of other nodes. Use of a broadcast based protocol allows the avionics to be implemented as a flat structure where nodes service the network rather than the other way around. With this design, aircraft control can be asserted by any of the Autopilot, Attitude Determination or Communications Interface modules (ground commands) or a combination thereof. Such an approach is in contrast to the more hierarchical approaches taken in the previous avionics design and in the design of avionics systems in general. With a distributed physical architecture more space can be made available for payloads. Wiring harnesses are simplified and production testing can be improved by installing diagnostics onto each node. Software implementation now becomes more solidly segregated, leading to greater robustness and a simpler system software qualification process. As a further gain, future changes to the hardware can be implemented sequentially rather than on an all or nothing basis. This both speeds time to market and ensures a better economic return over the life cycle of the product. Graceful degradation of the system now becomes possible with a distributed architecture. Previously, failure of the main processor in traditional designs would mean catastrophic failure of the aircraft (hence the normal requirement for redundancy). Because intelligence is now distributed throughout the aircraft under some conditions it would still be possible to fly the aircraft in a degraded state to a location where it could be safely ditched.