2022
Author: Alan Liu
Abstract:
Orthogonal Frequency Division Multiple Access (OFDMA) was introduced in the latest WiFi standard, IEEE 802.11ax, or WiFi 6, to improve the efficiency of a WLAN network. While OFDMA is well-defined for a single-hop WLAN topology, it is not very well defined for a multi-hop mesh network topology. The main problem when defining OFDMA for a mesh network is Resource Unit (RU) scheduling. When a group of subcarriers in a channel are allocated to a single user, that group of subcarriers is called an RU. This project explores different implementations of RU scheduling algorithms for a mesh network and attempts to define an RU scheduling algorithm to achieve fairness, which is defined by the similarity of the throughput of the first hop compared with that of the second and third hops, while also trying to achieve the lowest possible average latency between packets. Ns3, a network simulator, is then used to simulate OFDMA algorithms on a tree topology, which is used to model mesh networks. Specifying OFDMA for a mesh network is useful since mesh networks are a far cheaper alternative for internet infrastructure than wired networks, so improving the traffic flow in a mesh network for targeted applications like VoIP and video streaming would greatly improve the cost efficiency of internet access in a large building.
Author: Akshay Bhamidipati
Abstract:
With the rise of the metaverse, VR/XR (Virtual Reality/Extended Reality) as a technology has become significantly more prevalent. This has led to several paradigms in terms of distributing computing power with many modern VR concepts either performing all computation directly from the headset or engaging in offloading to an edge computer where data is streamed to the cloud, processed, and sent back. Adding another layer of granularity, streaming the data to the cloud can be done by either a wired connection such as an ethernet/HDMI cable or through a wireless link which, despite the slightly increased latency, eliminates problems such as limited mobility due to wires. Looking closer at the wireless approach, there are several more ways in which these networks can be configured, especially though the use of millimeter wave technology as opposed to Wi-Fi. Compared to traditional Wi-Fi, mmWave signals offer wider bandwidth and consequently greater throughput, in turn allowing for a more seamless virtual experience with high resolution XR feeds from the edge server at low latencies. We therefore explore a different direction for the streaming of XR traffic: utilizing emerging millimeter wave networks which introduce unprecedented bandwidth improvements that contribute to increased throughput and allow for streaming high resolution XR video. Focusing on achieving the best combination of low-latency and high throughput, we implement a network of one 60 GHz Mikrotik wAP and one 60×3 GHz Mikrotik wAP to simulate passing of data between the XR headset and offloading edge server.
2021
Author: Hailan Shanbhag
Abstract:
The push towards 5G networks has skyrocketed in the recent years. 5G aims to provide Gbps data rates, extremely low latency and more network capacity than was previously available. Millimeter wave (mmWave) technology provides a steppingstone for 5G networks because of its massive bandwidth from 30 GHz to 300 GHz. However, the attenuation of mmWave frequencies requires beamforming to focus the transmitted power toward the receiver. Because of this, mmWave radios use phased array technologies to alter and direct the radiation patterns. Larger phased arrays allow for even more focused beams. Unfortunately, commercial large phased arrays are not available in the market. In this project, we explore the process of combining multiple phased array modules to create larger phased arrays. Combining modules proves difficult because there are hardware discrepancies and limitations that affect the beam patterns. These include phase differences between antennas within a module and between modules. In addition, phase shifter resolution is limited by two-bit resolution which affects both the calibration process and the beam angular resolution. We combined four phased array modules from 60 GHz mmWave radios with 32 element phased array modules. Our results show that calibrating and combining phased array modules added the beam patterns constructively and created much narrower beams.
2020
Author: Zitong Chen
Abstract:
Recently, a handful of wireless communication protocols that are seeking fundamental improvement in spectrum utilization adopted OFDMA (Orthogonal Frequency Division Multiple Access) as the multiple access technique for data transmission due to its high efficiency and flexibility. This research provides a general simulation framework for OFDMA Uplink and Downlink data transmission and analyzes the performance of the system under different configurations and various channel conditions. Specifically, we present our investigation of the effect of signal-to-noise ration and symbol timing offset, and we gauge the performance in terms of SER (symbol error rate).
2019
Author: Luting Chen
Abstract:
The use of networking signals has been extended beyond communication to sensing, localization, robotics and autonomous systems in recent years. Specifically, recent advances in 5G millimeter wave (mmWave) have explored the possibility of expanding the use of mmWave beyond device communications and simple range sensing to a full-fledged imaging under low visibility conditions (fog, smog, snow, etc.). This thesis explores the use of mmWave imaging for humans, which could be incorporated into autonomous driving technology for pedestrian imaging in low visibility conditions. Unfortunately, certain challenges have been identified in mmWave imaging for humans, including its low resolution, the presence of fake artifacts resulting from multipath reflections, and the vibration of the human body. This thesis presents a system that can enable high-resolution mmWave imaging for humans that tries to address the above challenges by leveraging recent advances in deep learning, known as generative adversarial networks (GANs). In this thesis, we propose a GAN architecture that is customized to mmWave imaging and build a system that can significantly enhance the quality of mmWave images for humans.