a Lifetime Story of a 3-Year-Old Wireless Beacon System in ...

/ 15

From Conception to Retirement: a Lifetime Story of a 3-Year-Old Wireless Beacon System in the Wild

Yi Ding1,2, Ling Liu3, Yu Yang4, Yunhuai Liu5, Desheng Zhang4, Tian He2

Alibaba Group1, University of Minnesota2, Shanghai Jiao Tong University3, Rutgers University4, Peking University5

Heatmap of aBeacon in Shanghai City

1

A $107 Billion Market

3× Rider Sharing in 2019- ResearchAndMarkets.com

On-Demand Delivery (food/grocery)

/ 152

t0: t1:

Courier

t2: t3:

Customer

500K Stores in Multi-floor Malls Nationwide

Merchants

Delivery Process

/ 153

Accept Arrive Pick Up Deliver

t0: Accept t1: Arrive

Courier

t2: Pick Up t3: Deliver

CustomerMerchants

Why Accurate Arrival Time Matters?

Application Three : Order Progress Demonstration

(Where is my meal?)

Application Two: Overdue Order Investigation

(Late Arrival or Slow Preparation?)

Application One: Order Dispatching

(Localization and Time Estimation)/ 154

t1: Arrive

Courier

Merchants

Industry State-of-Arts

Solution One:Manual Report

Solution Two:Smartphone GPS

Reportable Area (~500m)

/ 155

Order Info.

Confirm Arrival

t1: Arrive

Academy State-of-Arts

Infrastructure-free:• Acoustic [MobiSys 11]• Magnetic [MobiSys 18-1]• IMU [MobiSys 18-2]• Electromagnetic [MobiCom 18]Limitations:• Need site survey and updating• Cannot be applied to dynamic

environment

Infrastructure-based:• Wi-Fi [NSDI 13, 16]• LED [MobiSys 2017]• RFID [NSDI 15]• QR Code [IoT Journal 2018]Limitations:• Deployment cost (All)• Need hardware modification (LED)• Restriction on smartphone hardware (RFID)• Need manual effort (QR Code).

/ 156

BLE Beacon• Easy to deploy• Accurate enough• Transparent to couriers• Acceptable cost (<$10 each)

aBeacon System – Alibaba Beacon System

Merchant

aBeacon Device

76kmShanghai

/ 157

Shop Courier

Alibaba Cloud

(0) Deploying & Binding

(2) Uploading

(3) Mapping

(1) Advertising

12,000+ shops

ID Tuple

110,000+ couriers

Deployment TimelineSc

ale

Stage 1:Conception

2018 2019 2020

# of Orders

# of Active Beacon Devices

/ 158

2017


ale

Stage 1:Conception

Stage 2:Customization

2018 2019 2020

# of Orders


/ 158

2017

Customized beacon devices:• Less cost ($8 each)• Longer lifetime (>=2 years)


ale

Stage 1:Conception

Stage 2:Customization

2018 2019 2020

Stage 3:Deployment

Stage 4:Operation

Server Issues

# of Orders

Spring Festival


BatteryRunout

(Retirement)Spring Festival& COVID-19

PVG

Spat

ial H

eatm

ap

2018/03

SHA75 k

m

(a) 2 Weeks into Deployment

2018/01 2019/09 2020/03

PVG

SHAPVG

SHA

PVG

SHA

(b) Fully Operational (c) 2 Months Before Battery Runout (d) Retired (Non-Operational)/ 158

2017

Performance: Utility (overdue ratio reduction)Utility(%)

0

1

2

3

4

FloorB2 B1 1 2 3 4 5 All

Impact of FloorObservation: aBeacon is more beneficial in

higher floors and basements

/ 159

Cumulative Gain = Cumulative Utility - Cost($

USD

) Empirical GainBreakeven

LineBreakeven Point(after 12 months)

Performance: Reliability (how many arrival events can be detected among all events?)

Reliability(P

i Reli)(%

)

0

25

50

75

100

Nokia

Motoro

laHuaw

eiLen

ovo

Sam

sung

OneP

lus

Apple

SHARP

Nex

us

iOSAllAndroid

Reliability(%)

75

80

85

90

Stayingduration(minutes)0 5 10 15 20 25 30 35

Impact of Staying Duration

Impact of Smartphone Hardware / 1510

Empirical Gain

Upper Bound Gain

Android Reliability

iOS Reliability

75% 85%

Performance: Reliability (how many arrival events can be detected among all events?)

Reliability(P

i Reli)(%

)

0

25

50

75

100

Nokia

Motoro

laHuaw

eiLen

ovo

Sam

sung

OneP

lus

Apple

SHARP

Nex

us

iOSAllAndroid

Reliability(%)

75

80

85

90

Stayingduration(minutes)0 5 10 15 20 25 30 35

Impact of Staying Duration

Impact of Smartphone Hardware

Lesson Learned: Reliability in the WildEven for arrival detection, the reliability is far from guaranteed in the wild due to multiple factors.

/ 1510

Android Reliability

iOS Reliability

75% 85%

Performance: Lifetime (the lifetime of each device)

Lesson Learned: Lifetime in the Wild• Battery may NOT be the major constraint for mobile/wireless devices,

since 40% devices survive longer than the environment (i.e., shops).

/ 1511

Shoplifetime(days)

0

365

730

Devicelifetime(days)

0 365 730

Shop closed Battery runout

2018 2019 2020


2017

Scal

e

System Evolution Reliability Lifetime

Lessons

Implication

Implication for Building Industrial Systems

/ 1512


Lessons Physical devices fail earlier than expected.

Implication Adopt existing devices.



Wireless beacon devices are NOT reliable (for regulation).

Implication Adopt existing devices. Hybrid solutions (BLE+GPS+Manual Report)



Wireless beacon devices are NOT reliable (for regulation).

Device lifetime are significantly affected by the environment.

Implication Adopt existing devices. Hybrid solutions (BLE+GPS+Manual Report)

Adaptive battery design.

Next Generation of aBeacon: aBeacon+

/ 1513

Shop Courier

Alibaba Cloud

(0) Deploying & Binding

(3) Mapping

(1) Advertising

ID Tuple

(2) Uploading

• No hardware or deployment cost.• No lifetime worries.• No battery worries.• Hybrid solution.

Merchant

Couriers

t0: Accept t1: Arrive t2: Pick Up t3: Deliver

Data Release

Permanent link: https://tianchi.aliyun.com/dataset/dataDetail?dataId=76359#

Data Size: 31,131 couriers at 2,466 shops in one month.

aBeacon Sensing Data

Courier Report Data

Courier Trajectory Data

/ 1514

https://tianchi.aliyun.com/dataset/dataDetail?dataId=76359

Thanks All Yi Ding [email protected], [email protected]

/ 1515 / 15

mailto:[email protected]

mailto:[email protected]

Tarzia, Stephen P., Peter A. Dinda, Robert P. Dick, and Gokhan Memik. "Indoor localization without infrastructure using the acoustic background spectrum." In ACM MobiSys, pp. 155-168. 2011.

Shen, Guobin, Zhuo Chen, Peichao Zhang, Thomas Moscibroda, and Yongguang Zhang. "Walkie-Markie: Indoor pathway mapping made easy." In 10th {USENIX} {NSDI} 13, pp. 85-98. 2013.

Adib, Fadel, Zachary Kabelac, and Dina Katabi. "Multi-person localization via {RF} body reflections." In 12th {USENIX} ({NSDI} 15), pp. 279-292. 2015.

Vasisht, Deepak, Swarun Kumar, and Dina Katabi. "Decimeter-level localization with a single WiFi access point." In 13th {USENIX} ({NSDI} 16), pp. 165-178. 2016.

Wei, Yu-Lin, Chang-Jung Huang, Hsin-Mu Tsai, and Kate Ching-Ju Lin. "Celli: Indoor positioning using polarized sweeping light beams." In ACM MobiSys, pp. 136-147. 2017.

Jeon, Kang Eun, James She, Perm Soonsawad, and Pai Chet Ng. "Ble beacons for internet of things applications: Survey, challenges, and opportunities." IEEE Internet of Things Journal 5, no. 2 (2018): 811-828.

Lu, Chris Xiaoxuan, Yang Li, Peijun Zhao, Changhao Chen, Linhai Xie, Hongkai Wen, Rui Tan, and Niki Trigoni. "Simultaneous localization and mapping with power network electromagnetic field." In ACM MobiCom pp. 607-622. 2018.

Shen, Sheng, Mahanth Gowda, and Romit Roy Choudhury. "Closing the gaps in inertial motion tracking." In Proceedings of the 24th ACM MobiSys, pp. 429-444. 2018.

Tian, Zhao, Yu-Lin Wei, Wei-Nin Chang, Xi Xiong, Changxi Zheng, Hsin-Mu Tsai, Kate Ching-Ju Lin, and Xia Zhou. "Augmenting indoor inertial tracking with polarized light." In ACM MobiSys, pp. 362-375. 2018.

References[MobiSys 11]

[NSDI 13]

[NSDI 15]

[NSDI 16]

[MobiSys 2017]

[IoT Journal 2018]

[MobiCom 18]

[MobiSys 18-1]

[MobiSys 18-2]

a Lifetime Story of a 3-Year-Old Wireless Beacon System in ...

Documents

Transcript of a Lifetime Story of a 3-Year-Old Wireless Beacon System in ...