Source-linked AI summary
Static Multipath Suppression in Power Delay Profile of Time-Varying Channel at 28 GHz
Jarosław Wojtuń, Bartosz Czaja, Tomasz Grasza, Jan M. Kelner, Cezary Ziółkowski, Radek Zavorka, Josef Vychodil, Tomas Mikulasek, Jiri Blumenstein, Ales Prokes, Aniruddha Chandra
TL;DR
Persistent paths from stationary objects can obscure weaker moving-scatterer contributions in time-varying radio channels. The paper compares temporal differencing and Doppler-notch filtering on wideband 28 GHz road measurements. All methods suppress persistent components and improve vehicle-reflection visibility, while filter responses affect RMS delay-spread distributions.
Problem
Strong persistent multipath can dominate received signals and obscure weaker time-varying contributions in environments with stationary infrastructure and moving objects.
Source: §I · p. 1Method
The study compares two temporal differencing filters and three Doppler-domain notch configurations using 28 GHz wideband measurements in a dynamic road environment.
Source: §I · p. 1Results
All methods attenuate stationary multipath and improve vehicle-reflection visibility, while filtering widens RMS delay-spread distributions and distinct frequency responses yield different distributions.
Source: §IV · p. 4, §V · p. 5Takeaways & Limitations
Static-multipath suppression matters when characterizing moving-scatterer propagation, and the selected filtering method should be considered when interpreting temporal-dispersion estimates.
Source: §V · p. 5
Abstract
from arXiv · showhide
This paper investigates static multipath suppression in a time-varying radio channel at 28 GHz using wideband measurements conducted in a dynamic road environment. Two temporal differencing filters and three Doppler-domain notch filtering configurations are compared in terms of their effects on power delay profiles (PDPs) and root mean square (RMS) delay spread. The results demonstrate that all investigated methods effectively suppress stationary multipath components, improving the visibility of vehicle-related reflections. However, differences in filter frequency responses influence the relative powers of dynamic multipath components and the resulting RMS delay spread distributions. The findings highlight the importance of static multipath suppression and filter selection in characterizing time-varying propagation channels.
I. INTRODUCTION
Persistent stationary paths can obscure weaker reflections from moving objects, complicating characterization of time-varying channels. This paper compares temporal differencing and Doppler filtering to suppress those paths and assess effects on temporal dispersion.
I. INTRODUCTION
- Persistent multipath from stationary infrastructure may dominate received signals and obscure weaker time-varying contributions in environments with moving objects.CIR, PDP, and RMS delay spread are used to describe temporal dispersion.Source: §I · p. 1
- Prior work discussed wideband channel sounding and temporal-dispersion estimation, measurement reliability, and verification using PDPs and derived parameters including RMS delay spread.The introduction positions separating persistent paths from moving-scatterer contributions as an additional challenge in dynamic scenarios.Source: §I · p. 1
- Temporal differencing and Doppler-domain filtering can suppress approximately stationary components, but their distinct frequency responses may affect the remaining dynamic multipath structure.Filtering may also modify moving-scatterer amplitudes and thereby affect PDPs and RMS delay spread.Source: §I · p. 1
- At 28 GHz, two temporal differencing filters and three Doppler-notch configurations are compared on the same dynamic-road measurements using PDPs and RMS delay spread.The study assesses stationary-component suppression and how filtering choice influences temporal-dispersion characterization.Source: §I · p. 1
A. Measurement Setup
The measurements used a synchronized, wideband mmWave sounder with horn antennas and repeated channel observations at 28 GHz.
B. Measurement Setup
- A Keysight vector signal generator and signal analyzer formed the Tx and Rx, with amplifiers and 5 m and 0.5 m low-loss coaxial cables in the chains.Both instruments support operation up to 44 GHz and 1 GHz instantaneous bandwidth; a GPS-disciplined reference provided time and frequency synchronization.Source: A. Measurement Setup · p. 2
- Identical horn antennas at both ends provided nominal 25 dBi gain over 26.5–40 GHz, with approximately 7° and 9° elevation and azimuth beamwidths.Source: A. Measurement Setup · p. 2
- The sounder used an 8 μs FMCW chirp with 1 GHz bandwidth and acquired 1,200 channel observations over 30 s at 28 GHz.Rx windows were triggered every 25 ms, giving a 40 Hz observation rate; each contained at least one complete chirp.Source: A. Measurement Setup · p. 2
B. Measurement Scenario
The road measurements combined persistent reflections from stationary roadside objects with transient, vehicle-related paths in a quasimonostatic, non-line-of-sight geometry.
B. Measurement Scenario
- Vehicle-related components were observed under non-line-of-sight conditions, with Tx about 5 m above the road and Rx about 1.5 m high near the roadside.The Tx–Rx separation was small relative to distances to observed vehicles, creating a quasimonostatic geometry; both antennas pointed toward approaching vehicles.Source: B. Measurement Scenario · p. 2
- Street furniture and metallic safety barriers produced persistent multipath, while vehicles entering, crossing, and leaving the observation region generated time-varying paths.Vehicle-path delays and received powers varied with vehicle position and movement.Source: B. Measurement Scenario · p. 2
- The Hradecká Street measurement scenario was used to investigate static-multipath suppression in PDPs and its influence on characterizing time-varying channel dispersion.The aerial view identifies Tx and Rx locations and the monitored traffic direction.Source: Figure 2 · p. 2, B. Measurement Scenario · p. 3
III. STATIC MULTIPATH SUPPRESSION METHODS
The methods target approximately stationary paths in dynamic environments, where static components may mask weaker reflections from passing vehicles.
III. STATIC MULTIPATH SUPPRESSION METHODS
- Two temporal differencing filters and a Doppler-domain notch filter were considered to suppress stationary multipath contributions in the estimated CIR.Source: §III · p. 3
- The methods address static paths that appear at approximately constant delays and may mask weaker vehicle reflections.Source: §III · p. 3
- The method comparison is framed around separating stationary and moving-scatterer contributions in dynamic propagation environments.Source: §III · p. 3
A. Temporal Differencing
Temporal differencing applies first- or second-order FIR filters to successive CIR observations to suppress approximately stationary multipath. The filters also affect dynamic components according to their temporal variation.
A. Temporal Differencing
- Both differencing filters eliminate ideally constant components, using FIR coefficients [1, –1] for first order and [1, –2,1] for second order.The first one or two observations, respectively, are excluded because the needed preceding samples are unavailable.Source: A. Temporal Differencing · p. 3
- Second-order differencing additionally suppresses components varying approximately linearly and attenuates slowly varying components more strongly.Source: A. Temporal Differencing · p. 3
- Both filters modify dynamic multipath amplitudes in a Doppler-dependent manner, so resulting PDP differences reflect their distinct frequency-selective characteristics.Source: A. Temporal Differencing · p. 3
B. Doppler-Domain Filtering
Doppler-domain filtering separates components by their temporal variation, applying a zero-centered notch to each delay bin’s Doppler spectrum. The notch width controls the suppressed frequency range before the filtered CIR is reconstructed and converted to a PDP.
B. Doppler-Domain Filtering
- For each delay bin, a DFT along observation time distinguishes static components expected near zero Doppler from moving-scatterer components generally at nonzero Doppler.Source: B. Doppler-Domain Filtering · p. 3
- The rectangular notch removes Doppler components within its half-width, after which inverse DFT reconstruction yields a filtered CIR for PDP calculation.Source: B. Doppler-Domain Filtering · p. 3
- The tested notch half-widths were 0.9, 1.4, and 1.9 Hz; increasing the width suppresses a progressively wider range of stationary and slowly varying components.Source: B. Doppler-Domain Filtering · p. 3
- Processing covered 30 s and 1,200 snapshots at 40 Hz, giving approximately 0.033 Hz Doppler-frequency resolution.This resolution allowed narrowband suppression around zero Doppler frequency.Source: B. Doppler-Domain Filtering · p. 3
C. Summary
Filtering is applied to the complex CIR before PDP calculation, and differing filter amplitude responses can affect both static suppression and dynamic multipath power.
C. Summary
- All filtering methods are applied to the complex CIR before the PDP is calculated.Source: C. Summary · p. 3
- Different filter amplitude responses can change stationary-component suppression and the relative power distribution of dynamic multipath contributions.Source: C. Summary · p. 3
- These response-dependent PDP differences should be considered when comparing RMS delay spread estimates.Source: C. Summary · p. 3
IV. RESULTS
All investigated filters attenuate stationary multipath and improve the visibility of vehicle-related trajectories, but their different frequency responses alter dynamic-path powers and RMS delay spread distributions.
IV. RESULTS
- All methods substantially attenuate stationary multipath, including a strong component near 0.7 μs, and the filtered PDPs make vehicle-related trajectories more visible.Figs. 4 and 5 show representative PDPs after first-order FIR and 0.9-Hz Doppler-notch filtering.Source: §IV · pp. 3–4, Figure 4 · p. 4, Figure 5 · p. 4
- FIR filtering can make some inclined traces, especially those from receding vehicles, appear stronger, whereas wider Doppler notches may suppress slowly varying moving-object components.The differences reflect the filters’ Doppler-dependent amplitude responses and their effects on dynamic multipath powers.Source: §IV · p. 4
- The unfiltered channel generally has lower RMS delay spread because strong stationary contributions concentrate received power, so its estimate combines stationary and dynamic propagation.Filtered trajectories show broadly similar time trends, with more pronounced differences around 8–11 s, 18–22 s, and 24–27 s.Source: §IV · p. 4
- All filtering methods produce wider RMS delay-spread distributions than the unfiltered channel, with the filtered interdecile range rising from 46.5 ns to 144.3–160.1 ns.Fig. 7 compares the empirical CDFs; Table I summarizes the percentiles.Source: §IV · pp. 4–5, Figure 7 · p. 5, Table I · p. 5
- Temporal FIR filtering yields median RMS delay spreads of 147.6 ns and 137.6 ns for first- and second-order filters, while Doppler notches yield 165.5–173.2 ns.The notch-filter medians depend on notch half-width.Source: §IV · pp. 4–5
- Filtering shifts the RMS delay-spread tails: P10 falls to 79.3–107.4 ns while P90 rises to 223.6–266.1 ns.The unfiltered channel has P10 = 151.7 ns, P50 = 165.0 ns, and P90 = 198.2 ns.Source: §IV · pp. 4–5
V. CONCLUSION
All investigated methods attenuated persistent multipath, improving vehicle-reflection visibility, while filter frequency responses influenced remaining component powers and RMS delay-spread distributions.
V. CONCLUSION
- The comparison evaluated two temporal differencing filters and three Doppler-domain notch filtering configurations using wideband measurements from a dynamic road environment.Source: §V · p. 5
- All investigated methods attenuated persistent multipath components, improving the visibility of reflections associated with moving vehicles.Source: §V · p. 5
- Static multipath substantially affects RMS delay-spread estimates: unfiltered PDPs show a narrower range than filtered profiles, whose distributions also vary with filter frequency response.The filtering methods produced broadly similar dynamic multipath structures, but differed in the relative powers of the remaining components.Source: §V · p. 5