James Martin

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2021-17

May 1, 2021

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2021/EECS-2021-17.pdf

From a system point of view, this body of work examines the design and scaling of directional wireless mesh networks. Effort has been made to look at both the physical and medium access layers of these systems and what can be done to improve the capacity of these networks so that they may perform efficiently at very large scales. This work takes a forward looking view of these networks. While much of this work is based around millimeter wave radios (especially 60 GHz), it is not limited by it. Many of the ideas are independent of frequency and therefore should apply at both lower and higher carrier frequencies. From the physical layer standpoint, I show that interference is a significant limitation to system capacity even for very large antenna arrays (1000s of elements), eliminating the idea of ‘pencil beams’ that do not cause or receive interference. Limiting and mitigating interference are investigated. To limit interference, I look at how the antenna array geometry may be altered and show that of the geometries I have examined, linear arrays perform best. To mitigate interference, I show that using receiver only adaptation and transmitter beam steering can be significantly better than using beam steering for both transmitter and receiver and helps to better approach the theoretical capacity limit. At the medium access layer, I present a new protocol called listen-only scheduling. The principle idea behind it is that nodes in a mesh network each independently determine their listening schedules without transmission scheduling. The independent scheduling is enabled by the use of independent transmitters and receivers and could be used at any carrier frequency. It is particularly attractive for directional mesh networks because of deafness and exposed node issues. I show that theoretically it should be able to perform at least within 20–30% of the maximum theoretical throughput. Lastly, future directions for this work are presented with some preliminary work. In particular, there are opportunities for cross-layer optimization between the physical, medium access, and network layer to avoid interference, increasing system capacity. One should be able to schedule in time using link management at the medium access layer and schedule in space and time by using routing at the network layer.

Advisors: John Wawrzynek

---

BibTeX citation:

@phdthesis{Martin:EECS-2021-17,
    Author= {Martin, James},
    Title= {Directional Wireless Mesh Network Design},
    School= {EECS Department, University of California, Berkeley},
    Year= {2021},
    Month= {May},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2021/EECS-2021-17.html},
    Number= {UCB/EECS-2021-17},
    Abstract= {From a system point of view, this body of work examines the design and scaling of directional wireless mesh networks. Effort has been made to look at both the physical and medium access layers of these systems and what can be done to improve the capacity of these networks so that they may perform efficiently at very large scales. This work takes a forward looking view of these networks. While much of this work is based around millimeter wave radios (especially 60 GHz), it is not limited by it. Many of the ideas are independent of frequency and therefore should apply at both lower and higher carrier frequencies.
From the physical layer standpoint, I show that interference is a significant limitation to system capacity even for very large antenna arrays (1000s of elements), eliminating the idea of ‘pencil beams’ that do not cause or receive interference. Limiting and mitigating interference are investigated. To limit interference, I look at how the antenna array geometry may be altered and show that of the geometries I have examined, linear arrays perform best. To mitigate interference, I show that using receiver only adaptation and transmitter beam steering can be significantly better than using beam steering for both transmitter and receiver and helps to better approach the theoretical capacity limit.
At the medium access layer, I present a new protocol called listen-only scheduling. The principle idea behind it is that nodes in a mesh network each independently determine their listening schedules without transmission scheduling. The independent scheduling is enabled by the use of independent transmitters and receivers and could be used at any carrier frequency. It is particularly attractive for directional mesh networks because of deafness and exposed node issues. I show that theoretically it should be able to perform at least within 20–30% of the maximum theoretical throughput.
Lastly, future directions for this work are presented with some preliminary work. In particular, there are opportunities for cross-layer optimization between the physical, medium access, and network layer to avoid interference, increasing system capacity. One should be able to schedule in time using link management at the medium access layer and schedule in space and time by using routing at the network layer.},
}

EndNote citation:

%0 Thesis
%A Martin, James 
%T Directional Wireless Mesh Network Design
%I EECS Department, University of California, Berkeley
%D 2021
%8 May 1
%@ UCB/EECS-2021-17
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2021/EECS-2021-17.html
%F Martin:EECS-2021-17