Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at 60 and 80 GHz
summary
The gist
The paper presents results from a vehicle-to-vehicle channel measurement campaign conducted in millimeter-wave (MMW) frequency bands at center frequencies of 60 GHz and 80 GHz…
This episode discusses
The paper
Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at 60 and 80 GHz · Read on arXiv
Radek Zavorka, Jiri Blumenstein, Tomas Mikulasek, Josef Vychodil, Hussein Hammoud, Wojtun Jarosław œřrasky, Aniruddha Chandra, ˉ Markus Hofer, Ŋ Jan M. Kelner, ˉ Cezary Ziołkowski, Christoph Mecklenbrauker Ψ Ales Prokes
Department of Radio Electronics, Brno University of Technology · University of Southern California · National Institute of Technology, Durgapur · Institute of Communications Systems, Military University of Technology · Institute of Telecommunications, TU Wien · Ŋ AIT Austrian Institute of Technology
This paper presents results from a vehicle-to-vehicle channel measurement campaign conducted in the millimeter-wave (MMW) frequency bands at center frequencies of 60GHz and 80GHz, each with a bandwidth of 2GHz. The measurements were performed in a dynamic oncoming-vehicle scenario using a time-domain channel sounder with high-resolution data acquisition. Power delay profiles were extracted to study the temporal evolution of multipath components, and the root mean square (RMS) delay spread was analyzed to characterize the temporal dispersion of the channel. The results demonstrate differences between the two frequency bands. At 60GHz, the RMS delay spread is well approximated by a Gaussian distribution with a higher median value, while at 80GHz it follows a lognormal distribution with a lower median. Furthermore, the number of resolvable multipath components was found to be nearly twice as high at 60,GHz compared to 80GHz, highlighting the impact of antenna beamwidth and frequency-dependent propagation mechanisms.
DOI: 10.1109/CAMA65664.2025.11335268
Transcript
Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.
Rosa: We're looking at the paper: "Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at 60 and 80 GHz".
Dev: Here's the gist of the paper. The paper presents results from a vehicle-to-vehicle channel measurement campaign conducted in millimeter-wave (MMW) frequency bands at center frequencies of 60 GHz and 80 GHz, each with a bandwidth of 2 GHz, demonstrating differences between the two frequency bands <ref:2603.17766#pg2> Measurement Campaign Description The measurement campaign focused on characterizing the high-mobility V2V channel in a real-world environment <ref:2603.17766#pg3> The measurement was conducted on the campus of the Faculty of Electrical Engineering and Communication (FEEC), Brno University of Technology (BUT), Brno, Czech Republic, in the area between the T12 building on Technicka Street, the sports complex, and the adjacent parking lot <ref:2603.17766#pg3> This location was selected because it combines open-space conditions with the presence of reflecting objects, creating a representative V2V environment <ref:2603.17766#pg3> The studied scenario considered two vehicles approaching each other from opposite directions <ref:2603.17766#pg4> Transmit and receive antennas with front-end units were mounted on the roofs of the vehicles, as illustrated in Fig.
Rosa: 2 <ref:2603.17766#pg5> The total channel impulse response (CIR) recording spanned 8 s, but for further processing only the most relevant 4 s segment with the highest signal-to-noise ratio (SNR) was selected <ref:2603.17766#pg3> Measurement Setup The transmit baseband section of the time-domain channel sounder was realized using the Xilinx Zynq UltraScale+ RFSoC ZCU111 platform <ref:2603.17766#pg6> High-speed digitalto analog converters (DAC) with a sampling rate of 6.144 GSPS generated the in phase (I) and quadrature (Q) components of the intermediate-frequency signal <ref:2603.17766#pg6> As a probing waveform, a frequency modulated continuous wave (FMCW) signal with both up- and down-chirp slopes was applied, ensuring a spectrum with minimal irregularities <ref:2603.17766#pg6> Upconversion to the target MMW bands was carried out using Sivers IMA converters, specifically the FC1005V/00 for 60 GHz and the FC1003E/03 for 80 GHz <ref:2603.17766#pg6> The resulting radio frequency (RF) signal was subsequently amplified by a QuinStar QPW-50662330-C1 amplifier at 60 GHz or a Filtronic Cerus 4 AA015 amplifier at 80 GHz, and then transmitted through an omnidirectional antenna <ref:2603.17766#pg6> Stochastic Channel Characterization The temporal dispersion of the radio channel is described by the RMS delay spread <ref:2603.17766#pg5> The RMS delay spread is defined as the square root of the second central moment of the PDP <ref:2603.17766#pg5> The statistical distribution of the RMS delay spread over the measurement interval is shown in Fig.
Dev: 6a, where the probability density function (PDF) for both frequency bands are depicted <ref:2603.17766#pg6> The results indicate that the two measured frequency bands exhibit distinct statistical behaviors <ref:2603.17766#pg6> The distribution of the RMS delay spread at 80 GHz is more accurately modeled by a lognormal distribution, with a corresponding median value of 16.71 ns <ref:2603.17766#pg6> Temporal Variation of Multipath Components The number of resolvable propagation paths over time t was extracted for both frequency bands <ref:2603.17766#pg6> The overall shape of the curves reflects the scenario of two vehicles passing in opposite directions <ref:2603.17766#pg6> The 80 GHz band is characterized by more pronounced short-term fluctuations around the moving average (red line), with a variance of approximately 12 paths, compared to about 5 paths at 60 GHz <ref:2603.17766#pg6> The maximum number of resolvable paths was nearly twice as high at 60 GHz, reaching about 80, whereas at 80 GHz it peaked at only about 40 <ref:2603.17766#pg6> The results confirm that the 60 GHz and 80 GHz frequency bands exhibit distinct statistical distributions of the RMS delay spread in the oncoming vehicle scenario <ref:2603.17766#pg6> At 80 GHz, the distribution follows a lognormal form with a lower median (16.71 ns), whereas at 60 GHz, the delay spread is well approximated by a Gaussian distribution with a higher median value (19.97 ns) <ref:2603.17766#pg6> These findings indicate that within the MMW spectrum, propagation characteristics cannot be generalized across frequency bands <ref:2603.17766#pg6> The analysis of the temporal variation of multipath components further shows that the richness and stability of the channel are strongly dependent on both frequency and antenna characteristics <ref:2603.
Title and authors: Dev: So, looking at the full title "Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at sixty and eighty GHz", it immediately tells us that the core contribution is comparing these two specific frequencies in a V2V context <ref:2603.17766#pg1,Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at 60 and 80 GHz>. It’s not just about measuring one band; it's about seeing how propagation fundamentally changes when you move from sixty GHz to eighty GHz <ref:2603.17766#pg1>.
Rosa: Right, and the authors are Jiri Blumenstein, Tomas Mikulasek, Josef Vychodil, Hussein Hammoud, Wojtun Jarosław Aniruddha Chandra, Markus Hofer, Jan M. Kelner and Cezary Ziołkowski. They’re coming from a lot of different institutions across Europe and the US.
Taro: I wonder if that diversity in where they are based influences their view on the results? For instance, if one group is more focused on theoretical modeling while another is focused purely on field measurements, how does that show up in the final comparison?
Dev: They are presenting analyses comparing the time-dependent statistical characterization of MPC propagating between two oncoming vehicles at both sixty gigahertz and eighty gigahertz <ref:2603.17766#pg1>. This means they aren't just looking at a single snapshot; they are analyzing how the channel changes over time.
Rosa: The main contributions listed in the paper are clearly three things: first, the analysis of MPC propagation between these two bands including their time-dependent statistical characterization; second, the evaluation and comparison of RMS delay spread characteristics and their statistical distributions at both frequencies; and third, an investigation into temporal evolution in number of propagation paths and evaluating differences between frequency bands.
Taro: The paper focuses heavily on the RMS delay spread statistics. That’s where I see a lot of practical implications for us when we think about latency in autonomous systems. Is that just a number, or does it dictate how fast a vehicle can react?
Dev: It’s more than just a number; the distribution matters because it tells you about the likelihood of getting certain delays. They show distinct statistical behaviors between the two bands in how they distribute this delay spread over time.
Rosa: And that distinction is key, because if you use a model based on one band to design a system for the other, you’re going to run into problems. The authors are stressing the importance of having accurate, frequency-specific channel models for designing and evaluating MMW vehicular communication systems.
The paper's summary: Rosa: To summarize what they actually did in this paper called "Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at sixty and eighty GHz", they conducted a measurement campaign to characterize the high-mobility V2V channel in a real environment <ref:2603.17766#pg1,Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at 60 and 80 GHz>.
Dev: They focused on getting data from that specific location—the campus of Brno University of Technology—which they chose because it has that mix of open space and reflecting objects, which is essential for a representative V2V setting.
Taro: So, they weren't just testing in a controlled lab environment; they were trying to capture the complexity of real-world reflections and scattering that you see on campus. That’s where I start getting interested in how this data translates to actual driving conditions.
Rosa: They studied two vehicles approaching each other from opposite directions, which creates that oncoming-vehicle scenario they need for this kind of analysis. The measurement spanned eight seconds, but they only used the most relevant four seconds with the highest signal-to-noise ratio to keep the data clean.
Dev: The methodology involved using that Xilinx Zynq UltraScale+ RFSoC platform, generating signals with a six point one four four gigasamples per second DAC, and applying frequency modulated continuous wave signals with both up and down chirp slopes as the probing waveform.
Rosa: Then they took that signal and upconverted it to the sixty GHz and eighty GHz bands using Sivers IMA converters, with specific components for each frequency <ref:2603.17766#pg1,60 GHz and 80 GHz>.
Taro: I’m thinking about how they managed the different hardware pathways—the different DACs, the two different upconverters, and then the specific amplifiers for each frequency—because managing that setup in a real-time loop is where things can go wrong.
Dev: They amplified the resulting RF signal using either a QuinStar QPW-fifty million six hundred sixty-two thousand three hundred thirty-C1 amplifier at sixty GHz or a Filtronic Cerus four AA015 amplifier at eighty GHz before transmitting it through an omnidirectional antenna <ref:2603.17766#pg3,a QuinStar QPW-50662330-C1 amplifier at 60 GHz or a Filtronic>.
The paper's improvements: Rosa: Now, moving on to what they suggest as improvements, the authors are pointing out that we need better AI systems trained to use these frequency-specific channel models derived from this research.
Dev: That means if you want an autonomous vehicle system to predict signal propagation accurately, it should be trained using these frequency-specific models so it can anticipate multipath effects depending on whether the communication happens at sixty GHz or eighty GHz <ref:2603.17766#pg1>.
Taro: I think that ties into my point about detecting those multi-path components better. The paper shows a clear difference in the number of resolvable paths over time, with the eighty GHz band having more pronounced short-term fluctuations, with a variance of about twelve paths compared to only about five paths at sixty GHz.
Rosa: That’s exactly what they are suggesting. AI algorithms should be enhanced to distinguish between resolvable and clutter multi-path components by learning from that empirical data in Figure seven which shows that difference in path counts.
Dev: From a controls engineer standpoint, this is interesting because it suggests we can build systems that adapt their processing based on the expected channel characteristics. You could switch to a strategy optimized for the Gaussian distribution at sixty GHz with its median delay spread of nineteen point nine seven ns versus something tailored for the lognormal distribution at eighty GHz with its lower median value of sixteen point seven one ns <ref:2603.17766#pg1>.
Taro: So, dynamically adjusting the communication protocol based on the RMS delay spread distribution—that means if you know whether you’re operating near sixty GHz or eighty GHz, you can choose a different strategy for your safety case <ref:2603.17766#pg1>. That makes sense for graceful recovery when things get messy.
Rosa: Exactly. It moves us from a one-size-fits-all approach to something much more tailored to the frequency band and the channel's actual behavior in that specific environment.
Conclusion: Dev: So, wrapping up on this paper, the main implication is that propagation characteristics in millimeter-wave are highly dependent on the frequency you choose. The authors confirm that we can’t generalize these propagation characteristics across different frequencies within the MMW spectrum.
Rosa: They summarize it by saying at eighty GHz, the distribution follows a lognormal form with a median of sixteen point seven one nanoseconds, while at sixty GHz, the delay spread is well approximated by a Gaussian distribution with a median value of nineteen point nine seven nanoseconds <ref:2603.17766#pg1>.
Taro: It’s interesting how different the numbers are—a difference between about twenty and sixteen nanoseconds for that median delay spread. That’s a significant difference in terms of potential latency for real-time control loops.
Dev: For the engineers, it means you need to design systems aware of this frequency dependency when planning those communication protocols, because one band requires a Gaussian model and the other requires a lognormal model.
Rosa: It shows that propagation characteristics aren't universal across these bands in this scenario. It’s about acknowledging that fact when we move into real deployment. The paper "Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at sixty and eighty GHz" provides concrete evidence for how to approach channel modeling in this area <ref:2603.17766#pg1,Vehicle-to-Vehicle Millimeter-Wave Channel Characterization at 60 and 80 GHz>.
Taro: I just think it confirms that the richer the channel is, the more complex it gets when you try to model it accurately. We still have a lot of work left on how to handle those complex scenarios gracefully and with robust recovery mechanisms when things go wrong.
Dev: Yeah, so that’s what we're taking away from this paper: frequency matters for the statistical distribution of the RMS delay spread in V2V communication. That’s the core message.
Rosa: And it sets a standard for how we should approach channel modeling when designing those systems based on MMW data.
More episodes
- Daily Summary for 2026-11-19
- 2111.07910-Mask-guided Spectral-wise Transformer for Efficient Hyperspectral Image Reconstruction
- 2203.02149-HDNet: High-resolution Dual-domain Learning for Spectral Compressive Imaging
- 2205.10102-Degradation-Aware Unfolding Half-Shuffle Transformer for Spectral Compressive Imaging
- 2311.10959-Structure-Aware Sparse-View X-ray 3D Reconstruction
- 2403.04116-Radiative Gaussian Splatting for Efficient X-ray Novel View Synthesis
- 2410.02572-Combining Pre- and Post-Demosaicking Noise Removal for RAW Video
- 2604.17453-Learned Nonlocal Feature Matching and Filtering for RAW Image Denoising
- 2603.20387-End-to-End Multi-Task Learning for Adjustable Joint Noise Reduction and Hearing Loss Compensation
- 2507.20825-Chirp-Permuted AFDM: A Versatile and Multifunctional Waveform for 6G