BlueFMCW: Random Frequency Hopping FMCW

• BlueFMCW is a frequency-hopping FMCW technique that employs pseudorandom tone selection for constant-envelope transmission and robust signal performance.
• It uses rigorous nonlinear optimization of hop probabilities and modulation indices to closely match target spectral shapes while preserving range and velocity resolution.
• Practical implementation leverages rapid frequency-switching hardware and lightweight phase alignment algorithms, making it ideal for radar and interference-resistant communications.

Random Frequency Hopping (BlueFMCW) refers to a class of frequency-modulated continuous-wave (FMCW) signaling schemes in which the instantaneous transmit frequency is rapidly and pseudorandomly switched among a set of discrete frequencies within a predetermined band. This approach achieves constant-envelope transmission, enables robust spectral shaping, dramatically reduces vulnerability to interference and spoofing, and maintains high-resolution sensing or communications performance. BlueFMCW integrates methodologies from both spread-spectrum communications and modern radar, and is characterized by rigorous statistical design of the hop parameters to realize bespoke power spectral densities and interference-mitigation properties (Callegari, 2014, Moon et al., 2020).

1. Fundamental Principles and Signal Model

BlueFMCW constructs its transmit waveform by partitioning the available frequency band into $N$ discrete tones, with hop frequencies $f_i = f_0 + \Delta f L_i$, where $f_0$ is the band center, $\Delta f$ controls total span, and $L_i\in[-1,+1]$ are normalized level indices. The baseband modulator $u(t)$ is a piecewise-constant (PAM) function updating every $T_c$ seconds as

$u(t) = L_i\quad\text{for } t\in[k T_c,(k+1)T_c)$

where each $L_i$ is selected randomly with probability $P_i$.

The transmitted BlueFMCW signal is

$$x(t) = \cos\phi(t),\qquad \phi(t) = 2\pi f_0 t + 2\pi \Delta f\int_{-\infty}^t u(\tau)d\tau$$

which guarantees continuous phase (constant envelope) across hops. This random hopping structure differs from classical linear FMCW chirps, as the start frequencies and hence the sequence of instantaneous frequencies are determined by a pseudorandom sequence, typically from a cryptographically strong source (Moon et al., 2020).

2. Power Spectral Density Synthesis and Optimization

A rigorous theoretical framework is established to match the output spectrum $S_x(f)$ to a target spectral shape $S_\text{target}(f)$. The power spectral density of the random FM waveform under i.i.d. hopping statistics is

$$S_x(f) = \sum_{i=1}^N K_1(L_i, f-f_0) P_i + \mathrm{Re}\left\{ \frac{[\sum_{i=1}^N K_2(L_i, f-f_0) P_i]^2} {1 - \sum_{i=1}^N K_3(L_i, f-f_0) P_i} \right\}$$

with kernels $K_{1,2,3}$ parameterized by the modulation index $m=\Delta f T_c$.

To enforce $S_x(f)\approx S_\text{target}(f)$, a nonlinear, constrained optimization is posed: $$\min_{\mathbf{P}\succeq 0,\,\sum_i P_i=1,\,m>0}\; J(\mathbf{P},m) = \int_{f_0-f_\gamma}^{f_0+f_\gamma} \lvert S_x(f; \mathbf{P}, m) - S_\text{target}(f)\rvert\,df$$ where $f_\gamma$ encompasses potential out-of-band leakage. The optimizer (e.g., SLSQP) searches over $\mathbf{P}$ and $m$; initialization strategies such as "spectrum slicing" around $L_i$ focus convergence to meaningful local optima. Small $m$ yields fast hopping and sparser active tone sets, while large $m$ trades hopping speed for denser spectral occupation (Callegari, 2014).

3. Hopping Strategy: Fading, Capacity, and Network Decentralization

Random frequency hopping dramatically enhances robustness in environments with interference and traffic uncertainty. In wireless networks, each user independently selects $v$ out of $u$ available sub-bands per transmission slot, varying the active sub-band set via a randomized FH strategy. For a decentralized system with $N$ users, the average per-user multiplexing gain at high SNR is

$M_i = \frac{1}{2} v_i \prod_{k\neq i}(1-v_k/u)$

yielding a sum multiplexing gain (SMG) of

$$\text{SMG}(v_1,\ldots,v_N) = \frac{1}{2}\sum_{i=1}^{N} v_i \prod_{k\neq i}(1-v_k/u)$$

For symmetric scenarios where all $v_i=v$,

$\text{SMG}(v,N) = \frac{1}{2} N v (1-v/u)^{N-1}$

The optimum hopping load is $v^*=u/N$. This enables BlueFMCW to maintain service capability and maximize spectrum utilization even when the number of active users $N$ is random and unknown, outperforming conventional frequency-division approaches under fluctuating loads (0911.5527).

4. Radar Applications and Interference/Spoofing Mitigation

In radar, BlueFMCW divides a conventional chirp into $N$ equal-duration sub-chirps. Each sub-chirp, indexed by $k$, uses a frequency

$f_{0,k} = f_c + (\sigma(k)-1)\Delta f$

determined by a pseudorandom permutation $\sigma$. The effect is that the instantaneous frequency trajectory "hops" at sub-chirp boundaries.

Key performance implications:

• The beat frequency for a true target is invariant to the hop order: $f_\mathrm{beat}^{(k)} = \alpha\tau + f_D$, independent of $f_{0,k}$.
• Coherent interference or spoofing energy collapses to a single bin in conventional FMCW but is uniformly scattered over $\sim N$ bins in BlueFMCW, reducing peak interference power by $\sim 10\log_{10} N$ dB.
• Range and velocity resolution are preserved (e.g., $\Delta R = c/2B$, $\Delta v = \lambda/2 T_\mathrm{CPI}$), provided that phase discontinuities at sub-chirp joins are corrected (Moon et al., 2020).

A lightweight phase alignment algorithm reorders beat segments via the inverse permutation $\sigma^{-1}$. This realigns the phase across sub-chirp boundaries, restoring the full bandwidth and signal coherence.

5. Spectrum Estimation, Compressive Sensing, and Robustness to Missing Data

For post-processing or spectrum monitoring (e.g., in ELINT or jamming), high-resolution time-frequency (TF) estimation of FH signals (including BlueFMCW) in the presence of missing observations or artifacts leverages bilinear TF representations and structure-aware Bayesian compressive sensing (BCS). After applying an exponential chirp-smoothing kernel and adaptive-optimal kernel in the ambiguity domain to preserve FH auto-terms and suppress impulsive cross-terms, the instantaneous autocorrelation function is mapped to the TF plane via inverse DFT. A hierarchical Bayesian model is used, with spike-and-slab priors exploiting neighborhood structures to promote horizontal (temporally contiguous) FH activity. Gibbs-sampling based inference yields robust spectrum estimation, achieving high accuracy even with up to $50\%$ missing data, vastly outperforming STFT and standard SBL/SLR approaches (Liu et al., 2018).

6. Practical Implementation and Design Considerations

Hardware requirements include a local oscillator or direct digital synthesizer capable of $<$100 ns frequency switching, now standard in modern RF CMOS. Digital receivers must buffer $N$ beat signals and reorder them; memory and computational costs are $O(N)$. The phase-alignment and FFT processing add minimal complexity over standard FMCW. Synchronization demands are unaltered. BlueFMCW is thus compatible with existing high-resolution FMCW hardware while offering significant spectral agility, interference immunity, and configurability (Moon et al., 2020).

Design Parameter	Function	Typical Range/Notes
$N$ (hop tones)	Resolution/interference scatter	8–128+
$m=\Delta f T_c$	Hopping rate/spectral shape	0.5–10
$\mathbf{P}$	pmf for hopping	Optimized
Oscillator agility	Frequency switch time	$<$100 ns

7. Applications and Performance Benchmarks

BlueFMCW is applicable to ultra-wideband radar for automotive safety, adversarial/jamming-resilient RF communications, EMC testing, and non-destructive ultrasonic inspection. In automotive radar simulation (24 GHz, $B=5.08$ GHz, $N=32$), BlueFMCW demonstrated median SIR of $\approx 19$ dB under dense interference, compared to $2$ dB for conventional FMCW. Phase alignment produced an additional $\sim 7$ dB SIR gain. Time-domain measurements agree with theoretical PSD predictions post-optimization (Moon et al., 2020, Callegari, 2014).

The underlying optimization methodology enables the construction of waveform libraries with tailored spectral characteristics and hop statistics, trading off hopping rate, spectral sparsity, and interference resilience to suit particular application scenarios.

---

References: (Callegari, 2014, 0911.5527, Moon et al., 2020, Liu et al., 2018)