Jingwen Bai ECE, Rice University Joint work with Chenxi Liu* and - - PowerPoint PPT Presentation

▶
jingwen bai ece rice university joint work with chenxi
SMART_READER_LITE
LIVE PREVIEW

Jingwen Bai ECE, Rice University Joint work with Chenxi Liu* and - - PowerPoint PPT Presentation

Jan 12, 2024 35 likes •502 views

Inc Increasing C sing Cellula llular C r Capa pacity U ity Using ISM sing ISM Band Side nd Side-c -cha hanne nnels: A ls: A F Fir irst Study st Study Jingwen Bai ECE, Rice University Joint work with Chenxi Liu* and Ashutosh

slide-1
SLIDE 1

Jingwen Bai

ECE, Rice University

Joint work with Chenxi Liu* and Ashutosh Sabharwal Rice University, *Tsinghua University

Inc Increasing C sing Cellula llular C r Capa pacity U ity Using ISM sing ISM Band Side nd Side-c

  • cha

hanne nnels: A ls: A F Fir irst Study st Study

slide-2
SLIDE 2

Before Smartphone Revolution

WiFi- Laptop Cellular- phone

slide-3
SLIDE 3

Today’s Smartphones

  • Cellular band:

– UMTS/HSPA+/DC-HSDPA (850, 900, 1700/2100, 1900, 2100 MHz); LTE (Bands 1, 2, 3, 4, 5, 8, 13, 17, 19, 20, 25) …

  • ISM band:

– 802.11a/b/g/n Wi-Fi (802.11n 2.4GHz and 5GHz) – Bluetooth 4.0 ...

slide-4
SLIDE 4

Use of ISM Band on Smartphones

  • Simultaneous use of ISM- and Cellular band

– Assisted GPS

  • Location accuracy

– Data Offloading

  • Cellular network congestion

– Data Forwarding

  • Wireless tethering
slide-5
SLIDE 5

WiFi Bluetooth ISM Side-channel! Radio

User Controlled Cellular Provider Controlled

Shared ISM ! band antenna

New Use of ISM Band: ISM Side-channel

  • Create side-channels for interference management to

increase the overall cellular network capacity

– Side-channels are established between mobile clients

  • Serve as an additional radio to access ISM bands when

available and controlled by cellular providers

– ISM bands are usually controlled by end-users

  • Local knowledge
  • Inefficient and unstructured

– Make centralized decision

slide-6
SLIDE 6
  • How often can we establish ISM side-channels

between smartphones?

  • PART I: Availability of ISM side-channels in highways
  • How can we benefit from ISM side-channels?
  • PART II: Impact on cellular capacity of future wireless

architecture (MU-MIMO and full-duplex network)

slide-7
SLIDE 7

Highway – WiFi Free Locations

  • No WiFi infrastructure

– Opportunity of using ISM band among users

Physical Traffic Congestion Dense Clusters: Side-channels Cellular Network Congestion

  • Rush hour: traffic congestion

– Need to invoke complicated techniques to increase cellular capacity

Resolve

slide-8
SLIDE 8
  • Highway: Practically no WiFi coverage
  • Methodology

– Measure WiFi channel strength between smartphones

  • Our designed Android Application
  • Range Test + Highway Traffic data = Estimate

– Use WiFi frequency band as an example in ISM band

PART I: Availability of ISM Side-channels in Highways

slide-9
SLIDE 9

Intra-vehicle and Inter-vehicle ISM Side- channels in Highway

Intra-vehicle Inter-vehicle

slide-10
SLIDE 10
  • Set up WiFi connection using WiFi Hotspot

– Measure the WiFi channel strength using our designed Android apps

Methodology

Client-side App Client Server Request to join Setup the connection Server-side App

slide-11
SLIDE 11
  • Set up WiFi connection using WiFi Hotspot

– Measure the WiFi channel strength using our designed Android apps

Methodology

Client-side App Client Server Request to join Setup the connection Server-side App

Entry Description

Time ¡ Timestamp ¡of ¡the ¡sampling ¡ WiFi ¡RSSI ¡ RSSI ¡of ¡the ¡Wifi ¡connec7on ¡ WiFi ¡SSID ¡ SSID ¡(name) ¡of ¡the ¡Wifi ¡connec7on ¡ La7tude ¡ La7tude ¡of ¡the ¡device ¡using ¡GPS ¡ Longitude ¡ Longitude ¡of ¡the ¡device ¡using ¡GPS ¡ Neighboring ¡ CellInfo ¡ Info ¡about ¡the ¡neighboring ¡cellular ¡network ¡ such ¡as ¡RSSI ¡and ¡Cell ¡ID ¡ ¡

Server-side and Client-side Logs

slide-12
SLIDE 12

Intra-vehicle Environment

  • For a compact car:

– Average RSSI is -34.5 dBm, with a standard deviation of 5.5 dBm.

slide-13
SLIDE 13

Mimic Intra-vehicle Environment

High-Scattering Indoor Environment

3 4 5 6 7 8 9 10

  • 70
  • 65
  • 60
  • 55
  • 50
  • 45
  • 40
  • 35
  • 30

RSSI (dBm) Distance (meters)

Average RSSI with standard deviation

  • High-scattering indoor

– Mimic vehicles of large size – SNR of 32 to 54 dB (assuming the WiFi noise floor is -95 dBm)

slide-14
SLIDE 14

Inter-vehicle Environment

  • Place two vehicles at different distance separation

– Measure inter-vehicle ISM side-channel

slide-15
SLIDE 15

Inter-vehicle Environment

10 20 30 40 50

  • 85
  • 80
  • 75
  • 70
  • 65
  • 60

RSSI (dBm) Distance (meters)

Average RSSI with standard deviation

Average SNR ≥ 15 dB

slide-16
SLIDE 16
  • Rush hour traffic counts on California State

– 900 California State Highways: interstate, CA Route, US Route – Calculate vehicle-to-vehicle range – Estimate of the smartphone-to-smartphone WiFi communication range

Rush Hour Highway Traffic Data

slide-17
SLIDE 17

V2V Range Histogram

0.05 0.1 0.15 0.2 0.25 0.3 0.35

Histogram

0 10 20 30 40 50 60 70 80 90 100 More

V2V (meter)

  • 69% of time, there is at least one ISM side-channel within 50m

– Given at least one smartphone per vehicle

slide-18
SLIDE 18

PART II: Impact on Cellular Capacity of Future Wireless Architecture

  • Trend of Base Station

– RF resources – Processing capability

Rice Argos Platform

Support multiple flows in the same cell simultaneously Massive MIMO BS Symmetric Traffic: MU-MIMO Asymmetric Traffic: Full-duplex Network

slide-19
SLIDE 19

MU-MIMO Full-duplex via ISM Side-channel Full-duplex Network MU-MIMO Downlink Performance Evaluation

PART II: Impact on Cellular Capacity of Future Wireless Architecture

slide-20
SLIDE 20

MU-MIMO Downlink

MU-MIMO ZFBF System with Perfect CSIT

  • Massive MIMO BS

– Zero-forcing beamforming (ZFBF)

  • Create orthogonal beam for each user

– With perfect Channel State Info at the Transmitter (CSIT), ZFBF can completely null out interference

slide-21
SLIDE 21
  • In practice: CSIT is not perfect

– Finite feedback bit à quantize channel instantiation

MU-MIMO Downlink

MU-MIMO ZFBF System with Imperfect CSIT

  • Imperfect CSIT à Inter-beam

Interference

slide-22
SLIDE 22

MU-MIMO Full-duplex via ISM Side-channel Full-duplex Network MU-MIMO Downlink Performance Evaluation

PART II: Impact on Cellular Capacity of Future Wireless Architecture

slide-23
SLIDE 23

Full-duplex Network

  • Asymmetric traffic: full-duplex doubles spectral efficiency
  • Massive MIMO BS

– Use some of the antennas for transmission and others for reception to enable full-duplex operation. – Passive self-interference suppression

  • Polarization, directionality, absorption

– Active self-interference cancellation

Bi-directional Full-duplex Full-duplex Network

slide-24
SLIDE 24

Total cancellation Mean 95dB+ Everett, Sahai, Sabharwal, Rice, 2013 Cancel Self-Interference Close to Noise Floor

Full-duplex Network

  • Mobile handsets remain half-duplex

Full-duplex Network with Half-duplex Clients

slide-25
SLIDE 25

Full-duplex Network

Full-duplex Network with Half-duplex Clients

  • Close distance between UL and DL

users à Uplink-downlink

Interference

slide-26
SLIDE 26

MU-MIMO Full-duplex via ISM Side-channel Full-duplex Network MU-MIMO Downlink Performance Evaluation

PART II: Impact on Cellular Capacity of Future Wireless Architecture

slide-27
SLIDE 27

MU-MIMO Downlink -- Inter-beam interference Full-duplex -- Uplink-downlink interference

MU-MIMO Full-duplex: Intra-cell Interference Crisis

Massive MIMO: MU-MIMO Full-duplex

slide-28
SLIDE 28

Improved Interference Management

  • Leverage ISM side-channels in dense environments
  • Our Solution

– Amplify-and-forward: Inter-beam interference for MU- MIMO downlink – Decode-and-cancel: Uplink-downlink interference for full- duplex network

slide-29
SLIDE 29

MU-MIMO Downlink: Amplify-and-Forward

  • A (B) amplifies the received signal and forwards it to B (A) on the

side-channel

  • A and B perform receive-beamforming to decode its own packet

based on all received signal and channel knowledge.

P a c k e t B Packet A

slide-30
SLIDE 30

Full-duplex: Decode-and-Cancel

  • C sends the packet encoded for the side-channel
  • B decodes Packet C, re-encodes, then cancels from main-channel
  • After canceling out Packet C, B can decode Packet B

Packet C Packet C

slide-31
SLIDE 31

MU-MIMO Full-duplex via ISM Side-channels

Base Station: M antennas BS can schedule the use of ISM side-channels for intra-cell interference management

  • BS: No CSIT is required

– Blindly serve DL with only K antennas – ZFBF to serve UL with the remaining antennas

  • DL users:

– amplify-and-forward

  • UL users:

– decode-and-cancel

slide-32
SLIDE 32

MU-MIMO Full-duplex via ISM Side-channel Full-duplex Network MU-MIMO Downlink Performance Evaluation

PART II: Impact on Cellular Capacity of Future Wireless Architecture

slide-33
SLIDE 33

Performance Evaluation

  • Goal: show the benefits of leveraging ISM side-channels

Area 50×50 square meters Base station antennas M = 20 Maximum number of users K + L = 20 Uplink and Downlink SNR 35 dB ISM Side-channel RSSI Refer to our measurement Main-channel Rayleigh fading

slide-34
SLIDE 34

Results I: MU-MIMO Downlink

  • Compare three systems with only downlink users:

– ZFBF with perfect CSIT

  • Use all M antennas

– ZFBF with finite-bit feedback

  • Use all M antennas
  • Finite-bit feedback:10 bits per user

– User cooperation via ISM side-channels

  • Amplify-and-forward
  • Use only K antennas without acquiring CSIT
slide-35
SLIDE 35

Results I: MU-MIMO Downlink

  • Compare three systems with only downlink users:

– ZFBF with perfect CSIT

  • Use all M antennas

– ZFBF with finite-bit feedback

  • Use all M antennas
  • Finite-bit feedback:10 bits per user

– User cooperation via ISM side-channels

  • Amplify-and-forward
  • Use only K antennas without acquiring CSIT
slide-36
SLIDE 36

MU-MIMO Downlink with Increasing User Density

slide-37
SLIDE 37

MU-MIMO Downlink with Increasing User Density

slide-38
SLIDE 38

MU-MIMO Downlink with Increasing User Density

slide-39
SLIDE 39

MU-MIMO Downlink with Increasing User Density

slide-40
SLIDE 40

Results II: MU-MIMO Full-duplex

  • Compare four systems with both up- & downlink users:

– TDMA

  • Serve one user at a time; Use all M antennas

– ZFBF downlink with perfect CSIT

  • Use all M antennas

– Full-duplex ZFBF with perfect CSIT

  • Use a subset of antennas for DL
  • Remaining antennas for UL

– MU-MIMO full-duplex via ISM side-channels

  • Use a subset of antennas to blindly serve DL without CSIT
  • Remaining antennas for UL
slide-41
SLIDE 41

Results II: MU-MIMO Full-duplex

  • Compare four systems with both up- & downlink users:

– TDMA

  • Serve one user at a time; Use all M antennas

– ZFBF downlink with perfect CSIT

  • Use all M antennas

– Full-duplex ZFBF with perfect CSIT

  • Use a subset of antennas for DL
  • Remaining antennas for UL

– MU-MIMO full-duplex via ISM side-channels

  • Use a subset of antennas to blindly serve DL without CSIT
  • Remaining antennas for UL
slide-42
SLIDE 42

MU-MIMO Full-duplex with Fixed User Density

slide-43
SLIDE 43

MU-MIMO Full-duplex with Fixed User Density

6.5X over ZFBF Downlink 12X over TDMA

slide-44
SLIDE 44

MU-MIMO Full-duplex with Fixed User Density

Recovering 2X full-duplex gain

slide-45
SLIDE 45

MU-MIMO Full-duplex with SNR

0.0 20.0 40.0 60.0 80.0 100.0 120.0 10 20 30 40 Expected Capacity (bps/Hz) SNR TDMA ZFBF downlink with perfect CSIT Full-duplex ZFBF with perfect CSIT MU-MIMO full-duplex via ISM side-channels

  • Expected capacity

– The expectation is taken over the estimate of the ISM side-channel range distribution we found in PART I

Ed ⇥ Capacity(d) ⇤

Expected Capacity Scales with SNR

slide-46
SLIDE 46

Conclusion

  • Availability of ISM side-channels

– 69% of time, we can establish ISM side-channels within 50m range on highway during rush hour with reliable link quality

  • ISM band for improved interference management

– Enable a flexible wireless architecture design of MU-MIMO full-duplex – Promise to improve the cellular network capacity many-fold