Tracking Space Debris Craig Benson School of Engineering and IT Enhancing space situational awareness using passive radar from space based emitters of opportunity
Space Debris as a Problem Debris is fast to stay in Low Earth Orbit requires around 8km.s -1 (28,800 km.hr -1 ) So even small objects have substantial kinetic energy Almost no drag, so even light objects stay in orbit for a long time Lots of it, and growing Iridium 33 & Cosmos 2251 Kessler syndrome proposes that there is a tipping point where collisions between debris cause more debris, which cause more collisions, and so on, with the potential to surround the Earth in a debris field that we cannot safely penetrate
Why should we care about Space Debris? Space provides substantial benefits: Communication and Broadcast satellites Remote Sensing and Weather satellites Imaging satellites Debris is a problem because: Risk of satellite loss due to collision Fuel burn to avoid collisions shortens satellite life Existing SSA is very poor Conjunctions are forecast with probabilities such as 10-6 The Iridium-Cosmos collision was not even the most likely at the time a. a. celestrak.com/events/collision.asp
Space Debris Tracking Lots of it Even small items are important impact can disable payloads Radiating items can be tracked by radio interferometers long baseline TDOA Debris doesn t radiate Non-functional satellites Rocket Bodies and fairings, lens covers, lost items Debris from previous collisions Optical observations are the main means of tracking nonradiating objects
Space Debris Optical Tracking At LEO the debris is illuminated in a dark sky for only short periods Each telescope is useful for only a few hours per day Lots of debris and sequential tracking means infrequent revisits Observations are only in 2 dimensions Largely infer altitude from rate of progress along track Radar would offer 24/7 opportunity
Radar Tracking of Space Debris Monostatic radar works a Space Fence works b Bistatic Radar works c a. Eastment, et al, Chilbolton presentation to UNSW SSA workshop 2014 b. Defence News, 2 June 2014. Lockheed was awarded a firm, fixed price contract of $914.7m for S-band (2-4GHz) array at Kwajalein c. Tingay, Kaplan, McKinley, el al, On the detection and tracking of space debris using the Murchison Widefield Array 2013
GNSS Bistatic Radar GNSS signals have unique properties for bistatic radar illumination from MEO (~26,600km) pseudo-random signal modulation L-band, which minimises atmospheric interaction very well monitored signal status around the world low cost receiving equipment reasonable effective aperture per omnidirectional element
GNSS Bistatic Radar GNSS has been used for passive radar before Initially by measurement of short-delay multi-path More recently to observe oceans from space
Passive Radar Concept
Passive Radar Concept Direct arrival of satellite signal is much stronger than signal scattered from debris Receive, Decode and recreate clean, phase perfect replica of illumination signal Use perfect replica of illumination signal
Passive Radar Concept 4,000m.s -1 8,000m.s -1
Passive Radar Concept Doppler 4,000m.s -1 8,000m.s -1
GNSS Bistatic Radar Link Budget for Space Debris From IS-GPS-200H Table 3.Va, S E rx -158.3dBW for L1 C/A From Grewal et al a SNR at typical GNSS receiver is approx -21.5dB a Grewal, Andrews, Bartone, GNSS, Inertial Navigation and Integration 3 rd ed. 2013
GNSS Bistatic Radar Link Budget for Space Debris! = 21.5!" +!! + 10!"#!" 4! + 2!!! +!!! Let R Dè R = 1000km RCS (re 1m2) SNR at typical receiver -20-21.5-20-11-120=-172.5dB -5.5-21.5-5.5-11-120=-158dB 10-21.5+10-11-120=-142.5dB From IS-GPS-200H Table 3.Va, S E rx -158.3dBW for L1 C/A From Grewal et al a SNR at typical GNSS receiver is approx -21.5dB a Grewal, Andrews, Bartone, GNSS, Inertial Navigation and Integration 3 rd ed. 2013
GNSS Bistatic Radar Link Budget for Space Debris Opportunities for improvement: Spatial Gain more elements in antenna bigger antenna Time integration (narrow bandwidth) 60dB in one second 78dB in one minute More emitters ~10 signals per satellite ~10 satellites per GNSS constellation
GNSS Bistatic Radar Link Budget for Space Debris Potential SNR from debris after Processing Gain 125dB = 1,000 elements, 7 minutes & 10 signals 173dB = 1,000,000 elements, 10 minutes & 300 signals SNR at typical receiver RCS (re 1m 2 ) G p =0dB G p =125dB G p =173dB -20-172.5dB -48.5-0.5-5.5-158dB -33 15 10-142.5-17.5 30.5
GNSS Bistatic Radar Challenges Variation in RCS Phase (target scintillation) Instability in Receiver Local Oscillator Non-predicted, Non-linear wander in Geometry (dynamic disturbance) Emitter Stability Ionospheric Models and Tracking Tropospheric Effects Observation duration (illumination angle) Using low-snr tracking outputs
Variation in RCS Phase (target scintillation) Radar target scintillation is due to complex scattering target >> wavelength Small debris is smaller than signal wavelength Babinet s principle applies Scattering is equivalent to diffraction through a hole shape and size of silhouette For a slot < wavelength, this is a high-school physics experiment, phase and amplitude are very well understood and predictable without exact target geometry GNSS signal wavelength is 19-25 cm Happy Place is for debris around 10cm in dimension Small enough to be predictable, not too small (Rayleigh scattering region)
GNSS Bistatic Radar Challenges Instability in Receiver Local Oscillator Non-predicted, Non-linear wander in Geometry (dynamic disturbance) Emitter Stability
Tracking GPS with a radio telescope a 13.8 13.75 phase offset from mean (cycles), arbitrary start point 13.7 13.65 13.6 13.55 13.5 13.45 13.4 13.35 13.3 2280 2290 2300 2310 2320 2330 2340 2350 2360 time (ms) RMS phase error to second order MMSE curve fit 17.6 a Data from ASKAP 15 Aug 2014 thanks to Dr Chris Phillips, CSIRO
Tracking GPS with a radio telescope a 13 phase offset from mean (cycles), arbitrary start point 12.9 12.8 12.7 12.6 12.5 12.4 6250 6255 6260 6265 6270 6275 6280 6285 6290 time (ms) RMS phase error to second order MMSE curve fit 17.4 a Data from ASKAP 15 Aug 2014 thanks to Dr Chris Phillips, CSIRO
GNSS Bistatic Radar Challenges Ionospheric Models and Tracking Tropospheric Effects Observation duration (illumination angle) Using low-snr tracking outputs repeated passes more data will refine astrodynamic models refining projected tracks rather than detecting random targets
Passive Radar Processing Concept Decode, store info, including data and carrier phase, recreate delayed and time distorted replica Path difference is of order 0-1500km, so delay of 0-5ms
Passive Radar Processing Concept If the coherent integration interval is long then multiple tracking hypothesis (tracking bins) may need to be tested Sensitive to shifts in carrier phase can only integrate coherently to +/-λ/4 Running a replica for each tracking bin scales poorly number of operations scales with N tracking bins Instead run shorter correlations in I & Q (sin and cos) store the complex (I&Q) correlations synthesise tracking bins later by rotating phases
Passive Radar Processing Concept I XOR XOR Q
Passive Radar Processing Concept Debris doppler Amplitude Actual debris signal T (100 s ms) Phase
Passive Radar Processing Concept Tracking bins are non-linear: FFT isn t directly useful But bins of interest can still be synthesised by twiddling phases but at linear cost per search (log cost for FFT) Very few bins are of interest must be (almost) energy conservative most are updates of known tracks search space = unmodelled disturbance
Passive Radar Processing Concept Early stage processing is 1-bit (XOR and up/down counter) Early stage processing is relevant to all bins for an object of interest Length of early stage processing limited by phase divergence of possible tracks Track synthesis performed as complex (phase and amplitude) rotations and additions of ~1-10k values
Passive Radar Processing Concept Debris Reflection GNSS Direct Signal Track Synthesis and Control Early Stage Process Early Stage Process
Conclusion Passive radar offers all-day, persistant surveillance energy efficient, multi-target, accurate tracking It is hard, but looks achievable