RFGo: A Seamless Self-checkout System for Apparel Stores Using RFID - - PowerPoint PPT Presentation

RFGo: A Seamless Self-checkout System for Apparel Stores Using RFID

Carlos Bocanegra (Northeastern University) Mohammad A. (Amir) Khojastepour (NEC Laboratories America) Mustafa Y. Arslan (NEC Laboratories America) Eugene Chai (NEC Laboratories America) Sampath Rangarajan (NEC Laboratories America) Kaushik R. Chowdhury (Northeastern University)

> <

AGENDA

2

Barcodes and Alternatives RFID as the key technology RFID-based proposals RFGo: vision, design and implementation RFGo evaluation Conclusions 6 5 4 1 2 3

> <

THE PAINFUL CHECKOUT PROCESS

3

[1] Forrester. 2018. Consumers Cringe At Slow Checkout. Forrester Opportunity Snapshot: Digimarc August 2018 (8 2018). https://www.digimarc.com/resources/forrester-study

13% sees the checkout burden as the decisive factor to switch to another store1 Only 23% of consumers are satisfied with the checkout process1

> <

ALTERNATIVES TO BARCODE

4

[1] Clresearch. 2018. ”Amazon Go and the Emergence of Sentient Buildings: How It Works and What Its Impact Will Be,” April 2018 (4 2018). http://www.clresearch.com/research/detail.cfm?guid=6A608036-3048-78A9-2FB3-4E6295D65919

Amazon Go: Cameras and sensors (sensor fusion) 1. Computer vision & Deep Learning Extensive hardware resources Dense camera deployment Privacy concerns 2. RFID based checkout

> <

RFID, WHY?

5

3. Governments embrace this technology, i.e. Japan 2025 initiative 1. Use cases of RFID in retail sector is on the rise

• Inventory management, Reduce out-of-stock items, Tracking items

at the warehouses 4. RFID is already in place for some major apparel retailers It is possible to build a seamless checkout system based on RFID

• it requires all items to be tagged

2. Cost per tag is low

> <

STATE-OF-THE-ART RFID CHECKOUT SYSTEMS

6

Handheld Cage-based Slot-based Bin-based Surface-based

Effortless Large area Unbarricaded Fast

> <

OUR VISION FOR A SELF-CHECKOUT SYSTEM

7

No manual effort Large checkout area High Speed checkout Unbarricaded checkout area

CA WA

Top view

EA

Customers and RFID items

CA

Checkout area

WA Wait

area

EA

Exit area

> <

BACKGROUND ON RFID

8

RFID basic components

CHIP And MEMORY ANTENNA QUERY

POWER

DATA

Backscattering

RFID TAG

ID ID

READER

R<->T communication phases

Center Frequency: 900 MHz Bandwidth: 1 MHz approximately Reader-To-Tag encoding: PIE Tag-To-Reader encoding: FM0

Gen2 configurations FM0 modulation

> <

CHALLENGES IN RFID

9

Blind Spots

❏ Illumination ❏ Orientation ❏ Coupling

P O W E R

Collision

RN16 R N 1 6 RN16

Position Uncertainty

❏ Mobility ❏ Non-stationary environment

> <

10

RFGo - OUR PROPOSED SYSTEM

1. Physical structure 2. Custom-built multi-antenna reader 3. Tag classification via supervised learning

Self-checkout vision

> <

11

RFGo - 1. PHYSICAL STRUCTURE

10 antennas, 6 covering the CA and 4 covering the outer region Unbarricaded and large CA with no need for manipulating the items IR sensors to assess occupancy within the CA Session: chain of operations including entry, scanning, classification and output

> <

• Conventional methods to resolve

collisions, effective but not in real-time

• RFGo - Exploits diversity in reception
• Multi-antenna commercial readers -

TDMA-TX/RX do not exploit diversity

RFGo - 2. CUSTOM-BUILT READER

12

Collisions

Low Reading Rate Slow checkout

> <

RFGo - 2. SELECTING THE RN16 TO ACK

13

• RN16 lack error detection mechanism, e.g.,

CRC

• SINR is a post-decoding metric and is not

available before decoding

• Can we find a pre-decoding metric which

follows the idea of SINR?

Packet Delivery Ratio (PDR)

Our solution: Interference Metric (IM)

> <

RFGo - 2. INTERFERENCE METRIC (IM)

14

FM0 symbols

20dB SNR 1 Tag

• Revisiting differential decoding

2 clusters (bits 1/0) 1 cluster

> <

10dB SNR 1 Tag

RFGo - 2. INTERFERENCE METRIC (IM)

15

30dB SNR / 5dB SIR 2 Tags 30dB SNR / 5dB SIR 3 Tags

Std/mean = 0.28 Std/mean = 0.33 Std/mean = 0.53

> <

RFGo - 2. INTERFERENCE METRIC (IM)

16

*IM -> IM Policy (IMP)

Packet Delivery Ratio (PDR)* SINR (Post decoding metric) IM (Pre decoding metric)

IM assesses the RN16 during decoding IM does not incur in extra computation cost IM is easily parallelizable across the RX-chains

> <

RFGo - 2. CUSTOM RFID READER IMPLEMENTATION

17

Smartrac Battery-less UHD RFID tags Octoclock for frequency And time sync USRP X310 with TwinRX daughterboards for RX-chains UBX daughterboard for single TX-chain Raspberry Pi controls the active TX antenna through MUX

[1] Nikos Kargas, Fanis Mavromatis, and Aggelos Bletsas. 2015. “Fully-Coherent reader with commodity SDR for Gen2 FM0 and computational RFID,” IEEE Wireless Communications Letters 4, 6 (2015), 617–620. https://doi.org/10.1109/LWC.2015.2475749

> <

18

RFGo - 3. TAG CLASSIFIER

Training stage uses a wide range of orientations and locations in the 3D plane

... ... ... ...

#readings_RX1 #readings_RX2 #readings_RXN RSSI_RX1 RSSI_RX2 RSSI_RXN Inside CA Outside CA

Neural Network formed by 3 hidden intermediate layers RSSI and # of readings as soft features for classification

> <

RESULTS - BENEFITS OF IM

19

IM impact on PDR

Packet Delivery Ratio (PDR) The fraction of slots that results in a correctly decoded EPC over the total number of slots.

• 6RX and 1TX. No blind spots.
• Slotted aloha saturates at 38%.
• FP reaches 52% resolving collisions
• Using the majority via MVP: 62%
• IMP wisely selects RN16 and

reaches 77% PDR.

> <

RESULTS - RECEIVER DIMENSION

20

IM impact on PDR and MDE

• IMP with 1TX and variable number of

RX antenna

• PDR increase from 50% to 73% with

6 antennas. It saturates after.

Packet Delivery Ratio (PDR) The fraction of slots that results in a correctly decoded EPC over the total number of slots.

> <

RESULTS - BENEFITS OF DISCOVERY RATE

21

IM impact on Discovery Rate

Discovery rate The percentage of the unique EPCs that have been decoded per unit of time.

Variable TX, 1 RX

• Multi-TX helps

dealing with Blind spots but is slow. Variable TX, 6 RX using IMP

• 2 RX helps speeding discovery by

almost 2x.

• 6RX achieve full discovery

under 1 second.

> <

Unified Cube

RESULTS - DEFINING THE CA AND GUARD AREA

22

The readings go 54’’ far from the CA. A classifier is needed

Untrained Inside-only features

Inside-only features considerable shrinkage of the spillover

Inside-only features

RFGo inside/outside features: spillover

• f 6 inches

> <

High precision when RFGo does not include an outside tag in the customer cart. High recall when RFGo detects all the items that is in the customer cart

RESULTS - PRECISSION AND RECALL

23

Deployment scenario

Horizontal Vertical Random

Experiment with Volunteers Recall of 99.68% Precision of 99.81%

872 tags

Experiment with multiple orientations Recall of 99.79% Precision of 99.77%

> <

• 1. RFGo, a first-of-its-kind self-checkout system based on RFID

CONCLUSIONS

24

• 2. RFGo enhances customer experience with its effortless, open and unrestricted

design

• 3. The multi-antenna framework increases the reading rate from 50% to 77%
• 4. The Supervised learning classifier achieves 99.79% precision and 99.77% recall

> <

THANKS

25