Real-time space debris monitoring with EISCAT

Following a feasibility study in 2000–2001 on using the EISCAT ionospheric research radars to detect centimetre-sized space debris in the frame of an ESA contract, we are now finishing a continuation study, aimed at achieving debris detection and parameter estimation in real-time. A requirement is to “piggy-back” space debris measurements on top of EISCAT’s normal ionospheric work, without interfering with that work, and to be able to handle about 500 h of measurements per year. We use a special digital receiver back-end in parallel with EISCAT’s standard receiver. We sample fast enough to correctly band-pass sample the EISCAT analog frequency band. To increase detection sensitivity, we use coherent pulse-to-pulse integration. The coherent integration is built-in in our method of parameter estimation, which we call the match function (MF) method. The method is derived from Bayesian statistical inversion, but reduces, with standard assumptions about noise and prior, to minimizing the least squares norm ∥z(t) ? bχ(R,v,a;t)∥, where z is the measured signal and {bχ} is a set of model signals. Because the model signals depend linearly on the amplitude b, it is sufficient to maximize the magnitude of the inner product (cross correlation) between z and χ, the amplitude estimate is then determined by direct computation. The magnitude of the inner product, when properly normalized, is the MF. To construct the set of model signals, we sample the EISCAT transmission, in the same way as we sample the received signal, and apply linearly changing Doppler-shifts to it. Our initial implementation of the MF-method in 2001 was about four orders of magnitude too slow for real-time applications, but we have now gained the required speed factors. A factor of ten comes from using faster computers, another factor of ten comes from coding our key algorithms in C instead of Matlab. The largest factor, typically 100–300, comes from using a special, approximative, but in practice quite sufficient, method of finding the MF maximum. Test measurements show that we get real-time speed already when using a single dual-processor 2 GHz G5 Macintosh to do the detection computations.