Choir: Empowering Low-Power Wide-Area Networks in Urban Settings - - PowerPoint PPT Presentation

▶
choir empowering low power wide area networks in urban
SMART_READER_LITE
LIVE PREVIEW

Choir: Empowering Low-Power Wide-Area Networks in Urban Settings - - PowerPoint PPT Presentation

Feb 17, 2024 411 likes •886 views

Choir: Empowering Low-Power Wide-Area Networks in Urban Settings Rashad Eletreby Diana Zhang, Swarun Kumar and Osman Yaan http://www.witechlab.com/LoRa/ChOIR.html 1 Imagine a world where every single object is connected to the Internet

slide-1
SLIDE 1

Choir: Empowering Low-Power Wide-Area Networks in Urban Settings

Rashad Eletreby

Diana Zhang, Swarun Kumar and Osman Yağan

1

http://www.witechlab.com/LoRa/ChOIR.html

slide-2
SLIDE 2

Imagine a world where every single

  • bject is connected to the Internet…

2

Few kbps Several miles away 10 year battery Simple and cheap RF interface

slide-3
SLIDE 3

3

Smart Infrastructure Smart Homes Smart Vehicles

slide-4
SLIDE 4

Low-Power Wide-Area Networking (LP-WAN)

slide-5
SLIDE 5

Long Range

  • Up to 10

KMs in rural areas Low Data rate

  • Order of

kilobits per second Low Cost

  • < $5

Low Power

  • Up to 10

years of battery life

Low-Power Wide-Area Networking (LP-WAN)

Initiatives from Industry (LoRa, SIGFOX) and standardization bodies (3GPP LTEM, NBIoT)

slide-6
SLIDE 6

Key Challenges

6

slide-7
SLIDE 7

Key Challenges

Interference

Collisions emerge from the sheer density of nodes and the simplicity of the current MAC protocols (e.g., transmit as soon as wakeup) LPWAN ranges drop by 10x in urban areas due to excessive multipath, shadowing, etc.

7

Range

slide-8
SLIDE 8

8

WiFi/Cellular Wireless sensor networks LPWANs

LoRaWAN Sigfox MegaMIMO SAM ZigZag Glossy ACR

Past work

…. …. ….

slide-9
SLIDE 9

Choir

Scalability

  • Decodes 10’s
  • f collided

transmissions Range

  • Extends the

range of teams of cooperating nodes Preserving simplicity

  • Fully

implemented at a single- antenna base station

9

base station over an area of 10 Km2 in Pittsburgh

slide-10
SLIDE 10

: Chirps

Chirp in T.D. Chirp on a spectrogram

Data encoding

10

slide-11
SLIDE 11

: 1-bit encoding

𝒐 bits -> divide the BW to 𝟑𝒐 initial frequencies

In general,

‘0’ ‘1’

11

slide-12
SLIDE 12

: 2-bit encoding

12

slide-13
SLIDE 13

: 2-bit encoding

13

slide-14
SLIDE 14

Choir in action

Interference

14

Range

slide-15
SLIDE 15

Collision of chirps

Different data

+

15

slide-16
SLIDE 16

Collision of chirps +

Same data

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

16

80 100 120 140 160 180

FFT Bin

100 200 300 400 500

  • Abs. FFT
slide-17
SLIDE 17

Hardware imperfections

𝑔 𝑔 + 𝜀𝑔

&

𝑔 + 𝜀𝑔

'

Local oscillator mismatch

17

slide-18
SLIDE 18

Hardware imperfections

Packet 1 Packet 2

Sub-symbol timing

  • ffsets

18

slide-19
SLIDE 19

19

Time Frequency TO Chirps are signals whose frequency increases linearly with time

slide-20
SLIDE 20

20

Time Frequency TO An offset in time maps to an offset in frequency! FO

slide-21
SLIDE 21

21

Time Frequency Two chirps with an offset in frequency! FO

slide-22
SLIDE 22

Collision of chirps +

Same data

80 100 120 140 160 180

FFT Bin

100 200 300 400 500

  • Abs. FFT

Hardware offsets!

22

slide-23
SLIDE 23

23

slide-24
SLIDE 24

Decoding data

U1 data: U2 data:

24

!

U1 data + U1 hardware offsets = 125 U2 data + U2 hardware offsets = 130

90 100 110 120 130 140 150 160 170

FFT Bin

100 200 300 400 500

  • Abs. FFT

Symbol 1

125 130

slide-25
SLIDE 25

Decoding data

Preamble

  • Sym. 1
  • Sym. 2
  • Sym. n

… Preamble

  • Sym. 1
  • Sym. 2
  • Sym. n

… Peak locations are used to estimate hardware offsets Hardware offsets remain constant across the packet

25

U1 data + U1 hardware offsets = 125 U2 data + U2 hardware offsets = 130

Symbol 1:

slide-26
SLIDE 26

Decoding data

Preamble

  • Sym. 1
  • Sym. 2
  • Sym. n

… Preamble

  • Sym. 1
  • Sym. 2
  • Sym. n

…

How to measure accurate hardware

  • ffsets across the preamble?

Peak locations are used to estimate hardware offsets Hardware offsets remain constant across the packet

26

slide-27
SLIDE 27

Decoding data

𝑔

& ∗, 𝑔 ' ∗ = 𝑏𝑠𝑕𝑛𝑗𝑜{3

4∈ 3467,3487 ,3 9∈ 3967,3987 } 𝑧𝐷6& − ℎ&𝑓?'@34A + ℎ'𝑓?'@39A

'

𝑔B

  • > initial frequency offset estimate of user i

ℎB

  • > channel estimate of user i

Δ

  • > bin size of the FFT

𝐷6& -> conjugate nominal chirp 𝑧

  • > received symbol

𝑔

B ∗

  • > correct frequency offset of user i

27

slide-28
SLIDE 28

28

Which peak corresponds to which user?

slide-29
SLIDE 29

29

Which peak corresponds to which user?

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

Symbol 2 27.2 189.6

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

Symbol 1 352.2 107.6

slide-30
SLIDE 30

30

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

Symbol 2 27.2 189.6

Which peak corresponds to which user?

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

Symbol 1 352.2 107.6

slide-31
SLIDE 31

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

Symbol 2 27.2 189.6

Which peak corresponds to which user?

100 200 300 400 500 600

FFT Bin

100 200 300 400 500 600

  • Abs. FFT

Symbol 1 352.2 107.6

31

User 1 User 2

Integer part depends on both data and hardware

  • ffsets

Fractional part depends

  • nly on hardware
  • ffsets
slide-32
SLIDE 32

1

Near-far effect

2

Inter-symbol interference

3

Handling a general number of collisions

32

We generalize this solution to account for…

slide-33
SLIDE 33

Choir in action

Interference

33

Range

slide-34
SLIDE 34

Range Extension

62 °F 67 °F 65 °F

Each node is out-of-range!

34

slide-35
SLIDE 35

Range Extension

62 °F 67 °F 65 °F

Each node is out-of-range!

35

63 °F 65 °F

Can we exploit data correlations to obtain a coarse- grained view of the sensed data?

slide-36
SLIDE 36

36 100 200 300 400 500 600

FFT Bin

5 10 15 20 25

  • Abs. FFT

65 °F 62 °F 67 °F

Noise floor

Objective

100 200 300 400 500 600

FFT Bin

5 10 15 20 25

  • Abs. FFT

64.5 °F

Noise floor

slide-37
SLIDE 37

37 100 200 300 400 500 600

FFT Bin

5 10 15 20 25

  • Abs. FFT

65 °F 62 °F 67 °F

Noise floor

Approach

100 200 300 400 500 600

FFT Bin

5 10 15 20 25

  • Abs. FFT

64.5 °F

Noise floor

Choir

Receive filter

slide-38
SLIDE 38

38 100 200 300 400 500 600

FFT Bin

5 10 15 20 25

  • Abs. FFT

65 °F 62 °F 67 °F

Noise floor

100 200 300 400 500 600

FFT Bin

5 10 15 20 25

  • Abs. FFT

64.5 °F

Noise floor

Choir

Receive filter

slide-39
SLIDE 39

Implementation

39

slide-40
SLIDE 40

Evaluation

40

slide-41
SLIDE 41

Hardware offsets

0.2 0.4 0.6 0.8 1 20 40 60 80 100 120 140

CDF Observed CFO+TO (Hz)

Observed Ideal

41

Hardware offsets are truly diverse across LPWAN radios

slide-42
SLIDE 42

Resolving interference

20000 40000 60000 80000 100000 120000 140000 2 3 4 5 6 7 8 9 10

Network Thrpt (bits/sec) # Users

Ideal

42

slide-43
SLIDE 43

Resolving interference

20000 40000 60000 80000 100000 120000 140000 2 3 4 5 6 7 8 9 10

Network Thrpt (bits/sec) # Users

Ideal ALOHA

43

slide-44
SLIDE 44

Resolving interference

20000 40000 60000 80000 100000 120000 140000 2 3 4 5 6 7 8 9 10

Network Thrpt (bits/sec) # Users

Ideal ALOHA ChOIR

44

29x

slide-45
SLIDE 45

Extending range

45

Number of collaborating nodes Range 1 1 Km 10 2.5 Km 30 2.65 Km 2.65X

slide-46
SLIDE 46

Conclusion

Objective Results

Scalability

  • Decodes 10’s
  • f collided

transmissions Range

  • Extends the

range of teams

  • f cooperating

nodes Preserving simplicity

  • Fully

implemented at a single- antenna base station

Platform

46