Intelligent Instruments for the Space Plasma Environment

The work of the Space Science Centre at the University of Sussex in the development of intelligent space plasma instruments is presented here. Previously the Centre has included various intelligent techniques within space instruments flown on a number of space missions. A neural network was included in the SPREE instruments flown on Shuttle flights STS-46 (1992), and STS-75 (1996). Fuzzy Logic control of telemetry compression and buffering was designed for the ELISMA instrument on MARS-96. Sussex pioneered the use of particle correlation via hardware and software processing as a means of studying plasma wave-particle interactions using particle detection pulses within particle sensors, (AMPTE UKS, CRRES, STS-46, STS-75, ESA Cluster II, and auroral sounding rockets). Large interacting arrays of microprocessors were employed to provide processing for the above activities (e.g. 20 separate processors were used within SPREE). Also fault-tolerant arrays of processors were designed for the MARS-96 ELISMA instrument. Current research at the Space Science Centre concentrates on the development of flexible space instruments compatible with on-board intelligence and on increased use of Field Programmable Gate Arrays, FPGA, for fast real-time implementations of dedicated complex algorithms. For example real-time plasma simulations of the spacecraft's plasma environment are being implemented in FPGA with local measurements used directly as input parameters. These simulations can then be used to optimise instantaneous instrument parameters and, most significantly, by comparing simulation results with actual measured parameters concentrate data transmission on phenomena whose physics is least understood. 1. ASPECTS OF INTELLIGENT INSTRUMENTS FLOWN TO DATE 1.1 Particle Correlation Since 1979 simple microprocessors have been added to energetic particle detectors so that fast, detailed information that would otherwise not be transmitted to ground could be processed on-board. Sussex pioneered the use of particle correlation as a means of studying space plasma wave-particle interactions (Gough , 1998a, Gough et al, 2002). Prior to the application of particle correlators electron detection pulses were simply counted as a function of direction of arrival and energy or velocity to generate averaged electron distribution functions. The fastest time resolution provided was typically 0.1s, usually limited by the energy cycle stepping sequence of the instrument. However, it was realised that electron detection times could be measured to nanosecond accuracy within the instrument and those arrival times would contain information about any resonant waves that the particles had interacted with in the space plasma local to the spacecraft. Fast electronics combined with microprocessors enabled Auto-Correlation Functions, ACF, to be generated from the fast particle detection pulses to cover all of the expected wave frequencies up to 10's of MHz. ACF were accumulated typically over seconds before transmission to ground, partly because of limited telemetry capacity and partly to improve measurement statistics. Particle correlators were flown on various auroral sounding rockets, AMPTE UKS, CRRES, STS-46, STS-75, and ESA Cluster II. The microprocessors involved were often only simple 8-bit microprocessors: National NSC800; radiation hardened Sandia 3000; and various Z80, 8031, 8051, etc. Despite the relative simplicity of application particle correlators furnished important new results about the nature of the interaction of natural auroral beams with the earth's ionosphere at auroral latitudes. Similarly man-made electron beams on Shuttle flights STS-46 and STS-75 were shown to interact with the ionosphere to generate a variety of waves from kHz to several MHz. Figure 1 illustrates some of the MHz electron modulations observed by particle correlators on STS-75 and highlights the ability to identify interacting wave type by observed dispersion, or variation of wave period with resonant electron velocity. Figure 1. Wave particle interactions observed by 10MHz particle correlators on STS-75. Electron ACFs with 64 lags of 50nS are plotted against 32 electron energy levels. Upper plot dispersionless modulations at electron gyrofrequency harmonics; Lower Plot dispersed modulations at frequencies corresponding to Bernstein waves. 1.2 On-Board Data Analysis by Neural Networks The SPREE electron and ion instrument flown on STS-46 (1992) and STS-75 (1996) was a complex instrument with many outputs, including data types corresponding to both normal particle counting and also to a large dataset of particle