Signal and noise in nonlinear devices
Independent stationary signal and noise processes \(s(t)\) and \(n(t)\) are the inputs of a nonlinear device whose output is the stationary process \(y(t)\). We write \(y(t) = s_y(t) + n_y(t)\), output signal plus output noise, where (i) \(s_y(t)\) is a function of the process \(s\), and (ii) \(n_y(t)\) is uncorrelated with all functions of \(s\). It follows immediately that \(s_y(t) = E(y(t) \vert s) + \text{const.}\), and we let \(\text{const.} = 0\). In the engineering literature, the covariance function of \(y\) has been split into signal and noise portions, but the processes \(s_y\) and \(n_y\) that generate these portions had not been identified. If the device is followed by a linear filter, the signal and noise portions of the final output are obtained by letting the filter act on \(s_y\) and \(n_y\), respectively. If the device is zero-memory, \(y(t) = F(s(t),n(t))\), then \(s_y(t)=G(s(t))\) where \(G\) depends on \(F\) and the distribution of \(n(t)\). If \(s\) is narrow-band, then \(G(s(t))\) can be expanded into harmonics. For the case of the hard limiter, \(F(s,n)=\text{sgn}(s+n)\), this leads to an easy derivation of signal-to-noise ratio in the various harmonic zones without using hyper-geometric functions. If the device is zero-memory, \(s\) is narrow-band, and \(n\) is Gaussian, the noise portion \(n_y\) can be written as a double sum of uncorrelated components, each of which is a harmonic of \(s\) times a Hermite function of \(n(t)\). The spectrum of this sum has appeared many times in the literature.
- Nonlinear devices acting as SNR amplifiers for a harmonic signal in noise
- A useful theorem for nonlinear devices having Gaussian inputs
- Hard-limiting of two signals in random noise
- On the output autocorrelation function of a single-valued nonlinearity with signal plus noise as input (Corresp.)
- The effect of a zero memory nonlinearity on the bandlimitedness of contaminated Gaussian inputs (Corresp.)
- The loss of signal detectability in band-pass limiters
- A useful theorem for nonlinear devices having Gaussian inputs
- The effect of instantaneous nonlinear devices on cross-correlation
- On the output autocorrelation function of a single-valued nonlinearity with signal plus noise as input (Corresp.)
- Performance of polarity correlators or limiters with pseudorandom additive input noise (Corresp.)
- Hard-limiting of two signals in random noise
- Theoretical analysis of the noise power ratio of nonlinear power amplifiers
- Nonlinear devices acting as SNR amplifiers for a harmonic signal in noise
This page was built for publication: Signal and noise in nonlinear devices
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2553338)