System Functions of the Channel

A Unified Framework for Channel Characterisation

We have encountered the time-variant impulse response h(Ο„;t)h(\tau; t) and the time-variant transfer function H(f;t)H(f; t). P. A. Bello (1963) showed that there are four equivalent system functions for an LTV channel, related by Fourier transforms. This elegant framework unifies delay-domain and Doppler-domain descriptions and leads to the scattering function β€” the most complete second-order characterisation of a WSSUS channel.

Historical Note: Bello's 1963 Paper

Philip Bello's 1963 paper "Characterization of Randomly Time-Variant Linear Channels" in IEEE Transactions on Communication Systems is one of the most cited papers in wireless communications. It established the mathematical framework that underlies all modern channel modelling, from 3GPP standards to 5G NR system design. The "Bello functions" remain the standard language of channel characterisation over 60 years later.

Definition:

Bello's Four System Functions

The four system functions and their Fourier relationships:

  1. Time-variant impulse response: h(Ο„;t)h(\tau; t) β€” response at time tt to impulse at delay Ο„\tau

  2. Time-variant transfer function: H(f;t)H(f; t) β€” frequency response at time tt H(f;t)=∫h(Ο„;t) eβˆ’j2Ο€fτ dΟ„H(f; t) = \int h(\tau; t)\, e^{-j2\pi f\tau}\, d\tau (FT over Ο„\tau)

  3. Doppler-spread function: S(Ο„;Ξ½)S(\tau; \nu) β€” spreading in delay Ο„\tau and Doppler Ξ½\nu S(Ο„;Ξ½)=∫h(Ο„;t) eβˆ’j2πνt dtS(\tau; \nu) = \int h(\tau; t)\, e^{-j2\pi \nu t}\, dt (FT over tt)

  4. Output Doppler-spread function: B(f;Ξ½)B(f; \nu) B(f;Ξ½)=∫H(f;t) eβˆ’j2πνt dtB(f; \nu) = \int H(f; t)\, e^{-j2\pi \nu t}\, dt (FT over tt), or equivalently FT of S(Ο„;Ξ½)S(\tau; \nu) over Ο„\tau.

These form a 2D Fourier transform pair: h(Ο„;t)↔τ↔fH(f;t)↔t↔νB(f;Ξ½)h(\tau; t) \xleftrightarrow{\tau \leftrightarrow f} H(f; t) \xleftrightarrow{t \leftrightarrow \nu} B(f; \nu) and h(Ο„;t)↔t↔νS(Ο„;Ξ½)↔τ↔fB(f;Ξ½)h(\tau; t) \xleftrightarrow{t \leftrightarrow \nu} S(\tau; \nu) \xleftrightarrow{\tau \leftrightarrow f} B(f; \nu).

,

Definition:

Scattering Function

Under the WSSUS assumption, the scattering function is

CS(Ο„;Ξ½)=E[S(Ο„;Ξ½) Sβˆ—(Ο„;Ξ½)]C_S(\tau; \nu) = E[S(\tau; \nu)\, S^*(\tau; \nu)]

It describes how the channel's power is distributed in the delay-Doppler plane. The scattering function provides the most complete second-order description of a WSSUS channel.

Its marginals are:

  • Integrating over Ξ½\nu: ∫CS(Ο„;Ξ½) dΞ½=P(Ο„)\int C_S(\tau; \nu)\, d\nu = P(\tau) (power delay profile)
  • Integrating over Ο„\tau: ∫CS(Ο„;Ξ½) dΟ„=SH(Ξ½)\int C_S(\tau; \nu)\, d\tau = S_H(\nu) (Doppler spectrum)
,

Bello's Four System Functions

Animated diagram showing the four system functions h(Ο„;t)h(\tau; t), H(f;t)H(f; t), S(Ο„;Ξ½)S(\tau; \nu), B(f;Ξ½)B(f; \nu) and their 2D Fourier transform relationships. The horizontal axis connects delay Ο„\tau to frequency ff; the vertical axis connects time tt to Doppler Ξ½\nu.
The four Bello functions form a 2D Fourier transform grid. Under WSSUS, the scattering function CS(Ο„;Ξ½)C_S(\tau;\nu) provides the complete second-order characterisation.

Scattering Function (Delay-Doppler)

3D surface plot of the scattering function in the delay-Doppler plane. Adjust the delay spread and Doppler spread to see how the channel energy spreads across both dimensions.

Parameters
300
100

Key Takeaway

The four Bello functions β€” h(Ο„;t)h(\tau; t), H(f;t)H(f; t), S(Ο„;Ξ½)S(\tau; \nu), B(f;Ξ½)B(f; \nu) β€” are connected by Fourier transforms in a 2D grid. Under WSSUS, the scattering function CS(Ο„;Ξ½)C_S(\tau; \nu) fully characterises the channel's second-order statistics, with the power delay profile and Doppler spectrum as its marginals.

Quick Check

The time-variant transfer function H(f;t)H(f; t) is related to the time-variant impulse response h(Ο„;t)h(\tau; t) by a Fourier transform over which variable?

Over tt (time)

Over Ο„\tau (delay)

Over both Ο„\tau and tt

Over ff (frequency)

Why This Matters: OTFS Modulation in the Delay-Doppler Domain

The scattering function CS(Ο„;Ξ½)C_S(\tau; \nu) lives in the delay-Doppler plane β€” the natural domain for doubly-selective channels. OTFS (Orthogonal Time Frequency Space) modulation places data symbols directly in the delay-Doppler domain using the symplectic finite Fourier transform, enabling each symbol to experience the full channel diversity. While OFDM operates in the time-frequency grid H(f;t)H(f; t), OTFS operates in the delay-Doppler grid S(Ο„;Ξ½)S(\tau; \nu), achieving better performance in high-mobility scenarios where the channel is sparse in delay-Doppler. The OTFS book develops this framework in depth, including connections to the Zak transform and practical receiver design.

Why This Matters: Channel Models for RF Imaging

In RF imaging (radar, SAR, near-field sensing), the channel impulse response h(Ο„;t)h(\tau; t) carries information about the scene reflectivity rather than being a nuisance to be equalised. The scattering function CS(Ο„;Ξ½)C_S(\tau; \nu) becomes the ambiguity function of the transmitted waveform convolved with the scene response. Delay corresponds to range, and Doppler to radial velocity β€” the same Bello framework from this chapter underpins target detection and imaging. The RFI book develops these connections, extending the channel model to include spatial (angular) dimensions for array-based imaging.

Scattering Function

CS(Ο„;Ξ½)C_S(\tau; \nu): power distribution in the delay-Doppler plane under WSSUS. Its marginals give the PDP and Doppler spectrum.

Related: Bello's 1963 Paper, WSSUS β€” Wide-Sense Stationary Uncorrelated Scattering, Delay Doppler

Doppler-Spread Function

S(Ο„;Ξ½)=Ft{h(Ο„;t)}S(\tau; \nu) = \mathcal{F}_t\{h(\tau; t)\}: describes channel spreading in delay and Doppler. One of Bello's four system functions.

Related: Bello's 1963 Paper, Scattering Function, Linear Time-Variant (LTV) System