How do I calculate the range resolution of an FMCW radar from its sweep bandwidth?
FMCW Range Resolution
Range resolution is arguably the most important radar performance parameter because it determines the radar's ability to separate closely spaced targets (objects, people, vehicles).
Beat Frequency Analysis
(1) The FMCW radar transmits a chirp signal: f(t) = f_0 + (BW/T_chirp) × t (linear frequency sweep from f_0 to f_0 + BW over duration T_chirp). (2) The received signal from a target at range R arrives at delay tau = 2R/c: f_rx(t) = f_0 + (BW/T_chirp) × (t - tau). (3) Mixing (multiplying) the TX and RX signals produces a beat frequency: f_beat = f_tx(t) - f_rx(t) = (BW/T_chirp) × tau = (BW/T_chirp) × (2R/c). (4) The beat frequency is digitized by the ADC and processed with an FFT. Each target produces a peak in the FFT at the corresponding beat frequency. (5) Two targets separated by Δr produce beat frequencies separated by: Δf_beat = (BW/T_chirp) × 2Δr/c. The FFT can resolve two frequencies if they are separated by at least 1/T_chirp (the FFT bin width). Setting Δf_beat ≥ 1/T_chirp: (BW/T_chirp) × 2Δr/c ≥ 1/T_chirp. Simplifying: Δr ≥ c / (2×BW). This is the range resolution, independent of T_chirp.
Improving Resolution
(1) Wider bandwidth = better resolution (the only way). No signal processing technique can improve the resolution beyond c/(2×BW) without additional bandwidth. (2) Synthetic bandwidth: transmit multiple chirps at different center frequencies and coherently combine the results. This effectively increases the total bandwidth beyond what a single chirp can achieve. Used in: stepped-frequency radar and some advanced automotive imaging radars. (3) Super-resolution algorithms (MUSIC, ESPRIT, CLEAN): these algorithms can estimate the position of point targets with accuracy better than c/(2×BW), but they do NOT improve the actual resolution (they cannot separate two closely spaced targets beyond the Rayleigh limit). Super-resolution works by exploiting the structure of the signal (knowing that the targets are point reflectors) and is not effective for complex, distributed targets (like a vehicle with multiple scattering centers). (4) Zero-padding the FFT: extends the FFT length beyond the number of samples, interpolating between frequency bins. This does NOT improve the resolution (the mainlobe width is unchanged). It does improve the accuracy of reading the peak position (better estimation of the target range, even though the two-target separation is unchanged).
f_beat = (BW/T_chirp)×(2R/c)
FFT resolution: 1/T_chirp Hz
4GHz BW: Δr = 3.75 cm
With Hann window: Δr ≈ 1.5×c/(2BW)
Frequently Asked Questions
Does the center frequency affect range resolution?
No. The range resolution depends only on the bandwidth: Δr = c/(2×BW). A 1 GHz chirp at 24 GHz has the same range resolution as a 1 GHz chirp at 77 GHz (both = 15 cm). However: the center frequency affects other parameters: angular resolution (higher frequency = better angular resolution for the same antenna size), maximum unambiguous velocity (proportional to lambda = c/f), and available bandwidth (the regulatory allocation at 77 GHz provides 4-5 GHz, while 24 GHz narrowband provides only 200 MHz). In practice: 77 GHz radars achieve better range resolution because more bandwidth is available (4 GHz vs 200 MHz), not because of the center frequency itself.
What bandwidth do I need for my application?
Application-specific requirements: (1) Automotive ACC (adaptive cruise control): need to track vehicles at 100-200 m range. Vehicles are separated by at least several meters. Δr = 0.5-1 m is adequate → BW = 150-300 MHz. But: modern ACC radars use higher BW (1-4 GHz) to also detect pedestrians and smaller objects. (2) Parking assist: need to detect obstacles at 0.5-5 m range. Objects can be 10-20 cm apart (parking bollard, curb edge). Δr = 5-10 cm → BW = 1.5-3 GHz. (3) Presence/gesture detection (60 GHz): need to detect hand features at 0.2-2 m. Finger separation ≈ 2 cm. Δr = 1-2 cm → BW = 7.5-15 GHz (achieved with V-band 57-71 GHz allocation). (4) Ground-penetrating radar: need 1-5 cm resolution in soil. Δr = 1-5 cm → BW = 3-15 GHz. Used at 1-10 GHz center frequency (lower frequencies penetrate soil better).
What limits the maximum bandwidth of a radar IC?
The VCO tuning range inside the radar IC limits the maximum chirp bandwidth. A VCO with a wider tuning range provides more bandwidth but is harder to design: (1) At 77 GHz: a VCO tuning from 76 to 81 GHz (5 GHz BW) requires a 6.5% tuning range. Achievable with modern SiGe or CMOS VCOs. (2) At 77 GHz with 7 GHz BW (76-83 GHz): requires 9% tuning range. More challenging; requires careful VCO design. (3) The PLL bandwidth and settling time must support the chirp rate. For a 50 us chirp with 4 GHz sweep: the frequency ramp rate is 80 THz/s. The PLL must track this ramp with < 0.1% error (< 4 MHz frequency error at any point). This requires a PLL bandwidth of at least 1-5 MHz. Additional limits: the PA must maintain flatness over the full bandwidth (gain variation < 1 dB), and the antenna must have sufficient bandwidth (the antenna VSWR must be < 2:1 over the full chirp bandwidth).