[Colloquium] Reminder - Muhammad Iqbal Rochman Dissertation Defense/Apr 1, 2024

[Megan Woodward]

This is an announcement of Muhammad Iqbal Rochman's Dissertation Defense.
===============================================
Candidate: Muhammad Iqbal Rochman

Date: Monday, April 01, 2024

Time:  9 am CT

Remote Location: https://uchicago.zoom.us/j/91496723089?pwd=SFZyaHc2WlY0SGwrYmJuRG54d1E4dz09  Meeting ID: 914 9672 3089 Passcode: 603129

Location: JCL 298

Title: Performance Evaluation of Wireless Networks Using Device-based Measurement

Abstract: Traditional methods for measuring cellular and wireless network performance, like simulations and lab experiments, often fail to capture the complexities of real-world deployments. Device manufacturers and network operators may face practical or economic limitations that prevent them from fully implementing network capabilities. This can include optional features like Multiple User MIMO (MU-MIMO) or massive MIMO, which may be omitted as long as basic device compatibility is maintained.  The resulting lack of clarity on actual network performance highlights the need for independent, academic measurement reports on deployed wireless networks, free from the influence of commercial operators. Our research investigates the design, approaches, and challenges of using consumer devices as viable tools for wireless performance measurement.  We explored methodologies for extracting data using both standard device Application Programming Interfaces (APIs) and root-access capabilities. Our findings demonstrate that this device-based methodology is both sufficient and scalable for characterizing previous-generation (4G and Wi-Fi 5/802.11ac) and current-generation (5G and Wi-Fi 6e) wireless networks, including their unique spectrum and feature sets.

Our methodology effectively measures the performance of newly deployed 5G networks in high-band/mmWave (>24 GHz) and mid-band (1-6 GHz), revealing crucial insights: (1) While we confirm the Gigabit-level downlink (DL) throughput of mmWave networks as advertised by operators, it suffers from limited range due to high-frequency signal fading. This necessitates dense and costly network infrastructure. Furthermore, indoor penetration, even with line-of-sight, is hampered by low-emission glass. Additionally, device thermal limitations can throttle throughput; and (2) The mid-band deployments, on the other hand, provide a balance of coverage and performance with a median of 200 Mbps DL throughput using a single 100~MHz channel, reaching 1~Gbps when aggregating multiple channels. This surpasses the median 31 Mbps achieved with a 20 MHz 4G mid-band channel. However, this improvement appears to be driven only by increased bandwidth availability. Moreover, we observe adjacent channel interference between 4G and 5G mid-band, likely due to insufficient guard bands and mismatched TDD configurations. These findings underscore the importance of additional spectrum for future 6G networks. However, operators and device manufacturers should emphasize the implementation of new 5G features like MU-MIMO and higher modulation for optimal performance.

Beyond cellular networks, our methodology has proven valuable in evaluating the newly released unlicensed 6 GHz spectrum utilized by Wi-Fi 6E. While current regulations mandate low-power indoor (LPI) deployments to protect incumbent fixed links, concerns persist regarding potential interference. Our extensive measurement campaigns across two university campuses indicate outdoor RSSI values between -64 and -95 dBm (median -89~dBm). Additionally, we observed building entry loss (BEL) of 12-16 dB through double-pane low-emission windows and 25-33 dB through solid brick walls. Notably, only 5% of indoor BSSIDs were detectable outdoors. These findings suggest a low risk of interference to incumbent fixed links. However, further research is warranted to determine appropriate signal levels for the proposed client-to-client (C2C) mode in future Wi-Fi 7 specifications.

Advisors: Monisha Ghosh

Committee Members: Heather Zheng, Nick Feamster, and Monisha Ghosh











-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20240401/cd9098ec/attachment.html>