Communication-centric integrated sensing and communications (ISAC) reuses data-bearing OFDM signals to sense targets. That reuse is efficient, but the range–Doppler map is no longer determined only by target delay, Doppler, and receiver noise. Random communication symbols and spatial beamforming also shape what the sensing processor sees.
A new public research record studies this effect when several MIMO data streams illuminate the same target. The central question is not merely whether more antennas improve sensing, but how multi-stream transmission changes the floor against which a target must be detected.
Multiple streams add a new matched-filter floor
The authors derive second-order moment expressions for the range–Doppler map under matched filtering and reciprocal filtering, then use those expressions to characterize dynamic range. Under matched filtering, the random superposition of several beamformed streams creates an additional data-induced floor.
This distinguishes the MIMO case from a familiar single-antenna result. With one stream, constant-modulus signaling can remove a modulation-dependent pedestal. With several spatial streams, constant modulus alone does not eliminate the extra floor produced by their random superposition.
Reciprocal filtering trades the data floor for noise amplification
Reciprocal filtering removes the matched-angle data-induced floor in the model, but it divides by the target illumination power. Weakly illuminated directions can therefore amplify receiver noise. The resulting floor depends on reciprocal-power statistics rather than only on the transmitted symbol distribution.
The two filters consequently favor different operating conditions. The public abstract reports that matched filtering is more robust under weak target illumination, while reciprocal filtering can deliver higher dynamic range when reciprocal-noise amplification remains mild.
Geometry becomes part of the sensing design
Because beamforming determines how user data streams illuminate the target angle, user–target angular geometry directly affects the filtering tradeoff. Sensing performance cannot be separated cleanly from the communication beam configuration in this setting.
The reported numerical results validate the derived analysis under the paper’s model. The public record does not establish how calibration error, clutter, waveform constraints, hardware nonidealities, or measured propagation would change the predicted floors in a deployed system.
Research notes
Delay-Doppler Sensing Performance Analysis for MIMO-OFDM ISAC Systems
Authors: Peishi Li, Rang Liu, Qian Liu, and Ming Li.
Status: Public arXiv record dated 1 September 2026.
What the public evidence establishes: The work derives second-order range–Doppler-map statistics for matched and reciprocal filtering, identifies an additional multi-stream floor under matched filtering, and characterizes a geometry-dependent dynamic-range tradeoff.
Limits: The available evidence is analytical and numerical; it does not establish performance in measured hardware, cluttered scenes, or channels outside the stated model.