The superheterodyne receiver
An echo at 3 or 10 GHz is hard to amplify and filter directly. Almost every radar (and every ESM receiver) therefore uses the superheterodyne principle: a mixer combines the echo with a local oscillator (LO) and converts it to a fixed, lower intermediate frequency (IF) — typically tens of MHz. The actual amplification and filtering happen at that IF, with components that perform far better than at microwave frequency.
That is the Achilles heel. A mixer cannot tell on which side of the LO a signal came in. A signal on the image frequency, 2 × IF away from the wanted frequency, therefore ends up in the IF just as easily. A filter in front of the mixer (the preselector) has to stop it; with double conversion — first to a high, then to a low IF — the image frequency lies so far away that filtering becomes easy.
Anyone who knows or can estimate a radar's LO frequency can jam on the image frequency — a frequency on which the radar itself does not operate and which a simple frequency agility strategy does not look at. Conversely, a receiver's LO leaks slightly to the antenna: a sensitive ESM receiver can sometimes pick up that leakage, even from a "silent" receiver.
Gain control, detection and I/Q
Behind the mixer, two mechanisms handle the enormous dynamic range between a distant, weak echo and strong clutter close in. STC (sensitivity time control) lowers the gain just after the transmit pulse and lets it rise with time, so that nearby clutter does not saturate the receiver. AGC (automatic gain control, Threshold, Constant False Alarm Rate and Automatic Gain Control) adapts the gain to the average signal level.
Then comes the detector. A simple amplitude detector gives only the envelope of the echo: enough for a non-coherent radar and a PPI picture, but the phase information is lost. A coherent radar uses an I/Q phase detector: it compares the IF signal with the COHO (Coherent amplifier chain) and with a copy shifted by 90°. That yields two components: I (in-phase) and Q (quadrature). Together they give the amplitude and the phase of every echo — and with that the raw material for Doppler filters (Pulse-Doppler radar and Moving Target Indicator), pulse compression and all digital processing.
From hits to plots and tracks
A radar delivers thousands of measurements per scan. The plot extractor turns them into targets, in three steps:
- Hit processor: for each range cell and each pulse it decides whether the echo exceeds the threshold (CFAR, Threshold, Constant False Alarm Rate and Automatic Gain Control). Every crossing is a hit.
- Sliding window: a window slides in azimuth over successive pulses and only declares a target once enough hits lie together — for example 8 out of 12 (an m-out-of-n criterion, see the hits per scan of Dwell time, hits per scan and integration). Isolated noise spikes drop out this way.
- Centre of mass: from the run of hits the bearing is determined at the centre of the group, more precisely than one beamwidth. The result is a plot: range, azimuth, time and amplitude.
Plots are snapshots. The tracker links plots from successive scans: it predicts where a known target should be now, places a gate (search window) around it and accepts the plot that falls inside. A new track is only started once a target appears consistently over several consecutive scans. Track-while-scan (Scan patterns of phased arrays) is this cycle, carried out while the radar keeps searching.
This is the chain that every false target (Deceptive jamming) has to survive. A deceptive jammer must not create a single spike, but a run of hits that gets through the sliding window, a plot in a plausible place, and scan after scan a kinematically credible motion — otherwise the tracker rejects it. Conversely, this explains why a range gate pull-off has to happen gradually: a jump falls outside the gate.
Radartutorial.eu: radar receivers and signal processing.