SmartVLC: When Smart Lighting Meets VLC Hongjia Wu 1 , Qing Wang 1 , - - PowerPoint PPT Presentation
SmartVLC: When Smart Lighting Meets VLC Hongjia Wu 1 , Qing Wang 1 , - - PowerPoint PPT Presentation
SmartVLC: When Smart Lighting Meets VLC Hongjia Wu 1 , Qing Wang 1 , 2 , Jie Xiong 3 , Marco Zuniga 1 1 TU Delft 2 KU Leuven 3 Singapore Management University Background - VLC Visible Light Communication (VLC): transmit data wirelessly using the
Background - VLC
Visible Light Communication (VLC): transmit data wirelessly using the artificial light (i.e., LED light) and is flickering-free for users.
SmartVLC 2/26
Background - Smart Lighting
Smart Lighting: the LED light changes with ambient light, to keep the total illumination constant within an area-of-interest more ambient light
- → less LED light
less ambient light
- → more LED light
SmartVLC 3/26
VLC and Smart Lighting in Action
Basic requirements
❼ Smart lighting: fine-grained dimming levels ❼ Visible Light Communication (VLC): optimized throughput
State-of-the-Art (SoA) solutions & Objective of this work LED’s brightness (dimming level) System throughput Strong ambient light Weak ambient light SoA-1 SoA-2 This work
SmartVLC 4/26
Related Work
SoA-1 (OOK-CT): On-Off-Keying (OOK) with Compensation Time
T
t
Data Compensation time
T
t
Data Compensation time
Dimming level: 40% Dimming level: 20%
SoA-2 (MPPM): Multiple Pulse-Position-Modulation
T
t
Choose 2 from 10: 45 possibilities
T
t
Choose 4 from 10: 210 possibilities Dimming level: 40% Dimming level: 20%
SmartVLC 5/26
Next Subsection
1 Motivation 2 SmartVLC
Design Implementation
3 Evaluation 4 Conclusion
SmartVLC 6/26
Basic Requirements
❼ Smart lighting: fine-grained dimming levels ❼ VLC: optimized throughput
SmartVLC 7/26
- 1. Support Fine-grained Resolutions
Increasing the N in MPPM does NOT work!
N ↑ ⇒ Symbol Error Rate ↑
Symbol: N1 time slots
à large N
…
Symbol: N2 time slots 0.2 0.4 0.6 0.8 1
Dimming level of the LED
2 4 6
Symbol Error Rate (SER)
10-3
N=10 N=30 N=50 N=80 N=120
SmartVLC 8/26
- 1. Support Fine-grained Resolutions
Proposed solution
Use multiplexing to achieve high resolutions.
t
25%
t
20%
t
30%
After multiplexing, we are able to capture all the brightness.
Before multiplexing (N = 10)
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Dimming level of the LED (N=10)
0.2 0.3 0.4 0.5 0.6 0.7
Normalized data rate
After multiplexing (N = 10)
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Dimming level of the LED (N=10)
0.2 0.3 0.4 0.5 0.6 0.7
Normalized data rate
SmartVLC 9/26
Basic Requirements
❼ Smart lighting: fine-grained dimming levels
SOLVED by multiplexing
❼ VLC: optimized throughput
SmartVLC 10/26
- 2. Increase System Throughput
Symbol definition
❼ S ∶< N,l >, N is number of slots, l is the brightness
e.g. S1 ∶< 10,20% >, S2 ∶< 10,30% > How about S1 ∶< 9,20% >, S2 ∶< 11,30% > ?
t
20%
t
30%
Obtain the dimming level in between of 20% and 30 %
SmartVLC 11/26
- 2. Increase System Throughput
Proposed solution
Adaptive Multiple Pulse-Position-Modulation (AMPPM): use super-symbols.
…
𝑇" 𝑂
",𝑚"
𝑇& 𝑂&,𝑚&
…
𝑇& 𝑂&,𝑚& Super-symbol: ⟨𝑇" 𝑂",𝑚" ,𝑛",𝑇& 𝑂&,𝑚& ,𝑛&⟩ 𝑇" 𝑂
",𝑚"
𝑢
Requirement: find the proper < N1,l1 > and < N2,l2 > to compose super-symbols for each required dimming level.
SmartVLC 12/26
- 2. Increase System Throughput
How to find < N1,l1 > and < N2,l2 >?
❼ Step 1: Collect available symbols 0.2 0.4 0.6 0.8 1
Dimming level of the LED
0.5 1 1.5 2 2.5
Symbol Error Rate (SER)
10-3 upper bound N=10 N=30 N=50
SmartVLC 13/26
- 2. Increase System Throughput
How to find < N1,l1 > and < N2,l2 >?
❼ Step 2: Calculate the normalized data rate for all the collected symbols
SmartVLC 14/26
- 2. Increase System Throughput
How to find < N1,l1 > and < N2,l2 >?
❼ Step 3: Obtain the optimal combination via multiplexing and slope-based selection.
0.5 0.55 0.6 0.65 0.7
Dimming level of the LED
0.7 0.75 0.8 0.85 0.88
Normalized data rate
N = 10 N = 11 N = 12 N = 13 N = 14 N = 15 N = 16 N = 17 N = 18 N = 19 N = 20 N = 21
without multiplexing AMPPM (with multiplexing)
S(21,0.571) S(21,0.524)
SmartVLC 15/26
Basic Requirements
❼ Smart lighting: fine-grained dimming levels
SOLVED by multiplexing
❼ VLC: optimized throughput
SOLVED by AMPPM
SmartVLC 16/26
Recap
LED light Intensity Time Compensation
- based approach
Compensation
- free approach
Our proposed approach Data Compensation Data (decoding unit) average Data (decoding unit) Data (decoding unit) Data (decoding unit) average average
SmartVLC 17/26
Next Subsection
1 Motivation 2 SmartVLC
Design Implementation
3 Evaluation 4 Conclusion
SmartVLC 18/26
- 3. Implementation
Available platforms
❼ High-end: USRP (≈ 5000✩), WARP (4900✩) ❼ Low-end: Arduino (≈20 e), BeagleBone Black (BBB, ≈65 e), Raspberry Pi (≈40 e)
BBB is selected in this project
❼ PRUs (similar to micro-controllers): BBB has two 200 MHz, 32-bit PRUs.
Used for sampling and operating LEDs.
Data Bridge
ARM PRU
SmartVLC 19/26
- 3. Implementation
Diagram of the front-end Transmitter and Receiver
Photodiode BBB MOSFET LED Power Ambient light Transmitter Photodiode Amplifier ADC BBB Receiver
Snapshot of the prototype Transmitter Receiver
SmartVLC 20/26
Next Subsection
1 Motivation 2 SmartVLC
Design Implementation
3 Evaluation 4 Conclusion
SmartVLC 21/26
Evaluation: Static Scenario
Setup: fix the window blind Results: throughput vs. distance
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Distance (m)
20 40 60 80 100 120
Throughput (Kbps)
dimming level=0.18 dimming level=0.5 dimming level=0.7
Observations
❼ Reliable communication within 3.6 meters ❼ Maximum throughput is ≈ 107 Kbps (dimming level = 0.5)
SmartVLC 22/26
Evaluation: Dynamic Scenario
Setup: lift the window blind Results: throughput vs. at a constant speed ambient light changes
10 20 30 40 50 60 70
Time (second)
50 60 70 80 90 100 110
Throughput (Kbps)
Ambient light: weak → strong
Observations
❼ The ambient light does not change averagely.
SmartVLC 23/26
Evaluation: Comparison
Comparison with OOK-CT and MPPM
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Dimming level of the LED
20 40 60 80 100 120
Throughput (Kbps)
AMPPM OOK-CT MPPM
Observations
❼ AMPPM outperforms OOK-CT (increases the throughput by 40% on average) ❼ AMPPM outperforms MPPM (provides non-flickering & increases the throughput by
12% on average)
SmartVLC 24/26
Conclusion
Co-design smart lighting and VLC:
1 AMPPM: fine-grained resolution & high throughput 2 Implementation: low-end platform1 3 Evaluation: static & dynamic scenarios 4 Others: heuristic coding scheme, brightness adaptation...
1Cited by OpenVLC 1.2 SmartVLC 25/26
Looking for PhD position!
SmartVLC 26/26