⚡ electrical engineering

A Preliminary Assessment of Midhaul Links at 140 GHz using Ray-Tracing

This paper evaluates the feasibility of using 140 GHz wireless links for 5G midhaul transport between Central and Distributed Units in an urban environment, concluding through ray-tracing simulations that this technology is promising for supporting multi-Gbps traffic requirements.

Original authors: Sravan Reddy Chintareddy, Marco Mezzavilla, Sundeep Rangan, Morteza Hashemi

Published 2026-05-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Sravan Reddy Chintareddy, Marco Mezzavilla, Sundeep Rangan, Morteza Hashemi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the internet as a massive, bustling city. In this city, the Central Units (CUs) are like the giant, powerful data centers (the "brain" of the network) located in the city center. The Distributed Units (DUs) are like the smaller cell towers on street corners or rooftops that actually talk to your phone (the "hands" of the network).

In the past, these two parts were connected by thick, expensive fiber-optic cables running underground. But as our data needs explode (think 5G and beyond), laying new cables everywhere is too slow and too costly. The researchers in this paper asked: "What if we could connect these data centers to the cell towers using invisible, super-fast wireless beams instead?"

They decided to test this idea using a very high frequency called 140 GHz (part of the Terahertz range). Think of this frequency as a "super-highway" that can carry a massive amount of traffic at once, but it's tricky to navigate because the signals are easily blocked by buildings.

Here is how they tackled the problem, using simple analogies:

1. The Challenge: Connecting the Dots

The researchers wanted to figure out the most efficient way to connect these towers. They had a map of a city area (Rosslyn, VA) with 8 potential data centers (CUs) and 36 cell towers (DUs).

The Goal: Use the fewest number of data centers possible to serve all the cell towers, while making sure every connection is fast enough to handle huge data loads.
The Problem: If you use too few data centers, the "wireless highway" gets jammed, and the speed drops. If you use too many, you waste money building unnecessary centers.

2. The Tool: A Digital Crystal Ball (Ray-Tracing)

Since building a real 140 GHz network in a city is incredibly expensive and requires hardware that barely exists yet, the researchers used a computer simulation called Ray-Tracing.

The Analogy: Imagine shining a flashlight in a dark room full of furniture. The light bounces off walls, tables, and chairs. Ray-tracing is like a super-smart computer program that simulates exactly how these "light beams" (radio waves) bounce, bounce, and bounce off buildings in a 3D city model. It tells them exactly which paths are clear and which are blocked.

3. The Strategy: The "Greedy" Matchmaker

To solve the puzzle of who connects to whom, they used a step-by-step method they call a "greedy algorithm."

The Analogy: Imagine you are a matchmaker trying to pair up 36 lonely dancers (DUs) with a few dance partners (CUs).
1. You look at all the potential partners and pick the one who can see the most dancers clearly (Line of Sight).
2. You pair that partner with the dancers they can see best.
3. You remove those dancers from the list.
4. You repeat the process with the remaining dancers until everyone has a partner.
  This method ensures they don't waste time trying to connect a dancer to a partner who can't see them.

4. The Secret Weapon: Giant Antenna Arrays

To make these high-speed connections work, the researchers assumed the data centers and cell towers would use massive arrays of antennas (16x16 grids).

The Analogy: Think of a regular Wi-Fi router as a lightbulb that shines light in all directions. The antennas in this study are like a laser pointer or a spotlight. They can focus the signal into a tight, powerful beam.
Because the signals are so focused, they can punch through the "noise" of the city and carry data at incredible speeds. The researchers used a technique called MU-MIMO (Multi-User MIMO), which is like a conductor directing a laser beam to hit multiple targets at once without the beams crashing into each other.

5. The Results: It Works!

After running their simulations, they found some encouraging results:

The Sweet Spot: They discovered that they didn't need all 8 data centers. Just 3 data centers were enough to serve all 36 cell towers.
The Speed: With these 3 centers, every single connection achieved a speed of 10 Gigabits per second. To put that in perspective, that's fast enough to download a high-definition movie in a fraction of a second.
The Condition: This worked best when the antennas were large (16x16) and the signals were focused correctly.

The Bottom Line

The paper concludes that using these super-high-speed 140 GHz wireless links to connect the "brain" of the network to the "hands" is a very promising idea. It suggests that in the future, we might not need to dig up streets to lay cables; instead, we can build a web of invisible, super-fast beams connecting our city's data centers, provided we use the right number of powerful antennas.

What they did NOT claim:

They did not say this technology is ready to buy today (hardware is still developing).
They did not claim this works for mobile phones moving in cars (they only tested stationary towers on rooftops).
They did not suggest this solves all internet problems, only the specific "midhaul" connection between data centers and cell towers.

Technical Summary: A Preliminary Assessment of Midhaul Links at 140 GHz using Ray-Tracing

Problem Statement
The exponential growth in mobile data traffic driven by 5G New Radio (5G-NR) and the anticipated demands of 6G networks place extreme capacity requirements on transport networks. To address this, the 3rd Generation Partnership Project (3GPP) has proposed functional splits that disaggregate the base station (gNB) into a Central Unit (CU), Distributed Unit (DU), and Remote Unit (RU). While this architecture centralizes processing, it creates a need for high-capacity midhaul links between CUs and DUs. In dense urban environments, deploying fiber for these links is costly. Although Terahertz (THz) communications (specifically above 100 GHz) offer the bandwidth and antenna density required for multi-Gbps rates, the feasibility of using 140 GHz specifically for CU-DU midhaul links in urban settings has not been fully explored. The core problem addressed is determining the minimum number of CUs required to serve a set of DUs in an urban environment while guaranteeing a specific peak data rate per link, utilizing 140 GHz wireless technology.

Methodology
The authors propose a system model and optimization framework to evaluate 140 GHz midhaul links in a 500 × 500 square meter urban area (Rosslyn, VA).

System Model: The scenario involves $C$ CUs and $D$ DUs. Each CU and DU is equipped with multiple antenna elements ( $N$ and $M$ , respectively, where $M \le N$ ). Paths between nodes are characterized by received power, Angle of Departure (AoD), Angle of Arrival (AoA), delay spread, and phase rotation.
Optimization Formulation: The objective is to minimize the subset of CUs ( $C'$ ) required to serve all DUs while ensuring each CU-DU link supports a peak rate $R$ (targeted at 10 Gbps). This is formulated as a binary optimization problem where variables determine CU selection and CU-DU association, subject to capacity and association constraints.
Algorithmic Approach: A greedy algorithm (Algorithm 1) is employed to select CUs and associate DUs. The algorithm iteratively selects CUs that maximize Line-of-Sight (LOS) paths to unassociated DUs. Once a subset of CUs is selected, DUs are associated based on the highest received signal strength.
Beamforming and Channel Estimation:
- Beamforming: DUs align their arrays with the strongest path. CUs, serving multiple DUs, approximate their boresight by averaging the directions of connected DUs.
- Channel Modeling: The channel matrix $H$ is computed as a linear combination of multipath components derived from ray-tracing data.
- Capacity Calculation: The system utilizes Space-Division Multiple Access (SDMA) and Downlink Multi-User MIMO (MU-MIMO). To manage interference without the complexity of optimal SINR maximization, the authors use Signal to Leakage plus Noise Ratio (SLNR) precoding. The achievable rate is calculated based on bandwidth, SINR, a loss factor, and a maximum spectral efficiency assumption.
Simulation Setup: Ray-tracing simulations were conducted using the Remcomm Wireless Insite tool with the X3D propagation model to capture channel parameters (path gains, delays, angles) in a 3D urban environment. These parameters were exported to MATLAB for MIMO channel matrix generation and capacity analysis.

Key Contributions

Feasibility Study: The paper provides a preliminary assessment of 140 GHz wireless midhaul links between CUs and DUs in a disaggregated 5G architecture, a specific use case not fully explored in prior literature.
Optimization Framework: It defines a method to minimize the number of required CUs while guaranteeing a target data rate, utilizing a greedy selection policy and SLNR-based precoding.
Integration of Concepts: The work unifies the concepts of 3GPP functional splits (specifically High Layer Split, Option 2) with THz communications to address transport network capacity.
Simulation Results: It demonstrates that with appropriate beamforming and MU-MIMO techniques, a small number of CUs can effectively serve a dense cluster of DUs at 140 GHz.

Results
Using a simulation setup with 8 potential CUs and 36 DUs in a 500 × 500 sq.m area:

CU Count: A single CU or two CUs were insufficient to meet the 10 Gbps peak rate requirement on all links. However, increasing the number of active CUs to 3 allowed all 36 DUs to be served with the required peak rate.
Performance Metrics: With 3 CUs and 16×16 antenna arrays at both ends, all links achieved a Signal-to-Interference-plus-Noise Ratio (SINR) of at least 19.5 dB, corresponding to the target 10 Gbps rate.
Antenna Configuration: The study observed that larger antenna arrays (16×16) significantly increased the number of links satisfying the peak rate constraint compared to smaller configurations (e.g., 8×8).
Power Allocation: The results were achieved assuming equal transmit power allocation among connected DUs.

Significance and Claims
The paper concludes that 140 GHz wireless midhaul links, when combined with 3GPP functional splits (specifically Option 2), represent a promising technology for transport networks in dense urban environments. The authors claim that their results demonstrate the feasibility of serving multiple DUs with a minimal number of CUs using THz frequencies. They position this work as a preliminary assessment, noting that while the current results are promising, future work will explore joint optimization of CU selection and power allocation, more advanced beamforming techniques, and clustering algorithms to further improve spectral efficiency and data rates. The paper does not claim to have solved all challenges of THz deployment but establishes a foundational proof-of-concept for the specific CU-DU midhaul scenario.