1 of 61

CSE 122/222C ; WES 269�BLE Advertising Deep Dive

Pat Pannuto, UC San Diego

ppannuto@ucsd.edu

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

2 of 61

BLE Advertising Deep Dive Goals

Refresh Major Concepts in Bluetooth

Investigate some questions of real-world performance!

What are real-world use cases of advertisements?

How much energy do advertisements take?

What is the probability of receiving a packet?

Of receiving data?

2

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

3 of 61

Outline

Recap

Communicating with Advertisements

Advertisement Use Cases
Energy Use
Packet Collisions

3

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

4 of 61

Bluetooth Classic and BLE co-exist & have different use cases

4

Car BT Audio

Bluetooth Low Energy (introduced in v4.0)

Small amounts of data transfer
Low power requirements

“months or years” on small batteries

Ideal for IoT devices

Bluetooth Classic (around since 1999)

Continuous data communication
Can afford higher battery power requirements

Headphones

Mobile Photo

Printers

Fitness Trackers

Tracking Devices�(eg: Apple AirTag)

Switches

and Sensors

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

5 of 61

Bluetooth Low Energy Roles: Scanner, Advertiser

5

Scanner

“Current temp is 76F"

Advertiser

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

6 of 61

BLE Central and Peripheral roles are for connections

6

Central

“I’m the Living Room Light”

Peripheral

Connection Request

Connected (stops advertising)

“Set light brightness to 50%”

(note: this is highly simplified)

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

7 of 61

BLE Layers

Host – Configuration for advertisements, connections

GAP – Generic Access Profile

Configure advertising

GATT – Generic ATTribute profile

Configure connections

HCI - Host Controller Interface

Controller - Communication

Link Layer – send packets
RF and PHY – send bits

7

GAP

Host

Controller

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

8 of 61

What’s the need for the HCI Layer?

8

Host layers run

on computer

Controller layers run

on BLE hardware

HCI can be software API,

or physical (USB, RS232,..)

BT Classic envisioned separate implementations for Host and Controller

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

9 of 61

Advertisement and Scanning in Action

Advertisements are received when the channel of the scan window and the channel of the advertisement overlap in time (and space)

9

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

10 of 61

Advertisement payloads are in Length – Type – Value Format

Scanner hops through length/type pairs to find what interests it

10

Name
Service UUID
TX Power Level
Manufacturer-specific data
And about twenty others
Flags

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

11 of 61

Outline

Recap

Communicating with Advertisements

Advertisement Use Cases
Energy Use
Packet Collisions

11

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

12 of 61

Advertisements alone can be used for communication

Recall: BLE advertisements are uncoordinated, broadcast messages designed for discovery

Yet some devices, applications use only advertisements:

Beacons – iBeacon, Eddystone
Tracking – Tile, Apple AirTag
Local communication – Apple Continuity
Sensor deployments – PowerBlade

12

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

13 of 61

Beacons

Advertising with advertisements!

Web of Things

Real-world tags that broadcast virtual-world identifiers

iBeacon and Eddystone

Formats for sending URLs and device identifiers
Use existing BLE fields (Service Data and Manufacturer-Specific Data)

13

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

14 of 61

Beacons to advertise presence and send out useful information

iBeacon and Eddystone

Popular formats for sending URLs and device identifiers
Use existing BLE fields (Service Data and Manufacturer-Specific Data)

14

Self-guided tours at museums

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

15 of 61

Tracking

Find devices nearby

Get a sense of distance to the device

Find my X

Tile: find my keys
Apple: find my device

Uses TX power level field

Lists the transmitted power of the device
Pathloss = TX power – RSSI (all in dBm)

15

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

16 of 61

BLE signals can be used to track distance to items

Find my X

Tile, AirTag: find my keys
Apple: find my device

Find devices nearby

Get a sense of distance to the device

16

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

17 of 61

nRF Connect Activity

Work in pairs with your neighbor
Open nRF Connect app on both smartphones
1 person becomes Advertiser
Other person becomes Scanner

Advertiser:

Go to Peripheral
Set a name for your Advertiser
Start Advertising

Scanner:

Start scan (Play button on top right)
Filter by Advertiser’s name
Go to RSSI viewer

17

Advertiser

Scanner

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

18 of 61

RSSI and some questions

RSSI (Received Signal Strength Indicator): Strength of beacon’s signal as seen on receiving device

What’s the highest RSSI value you notice?�How close to the advertiser is this?

What do you notice about RSSI when moving away from the advertiser?

18

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

19 of 61

RSSI Experiments

Can you get different RSSI readings at the same distance?
Can get the same RSSI readings at different distances?

19

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

20 of 61

Intuition on using signal strength to find distance

RSSI (Received Signal Strength Indicator): Strength of beacon’s signal as seen on receiving device

20

Measured Power: -20 dBm

factory-calibrated, read-only constant

“Hey you should be getting an RSSI of -20 dBm at 1 m distance from me”�

RSSI from

received signal

Can roughly figure out

how far the beacon is

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

21 of 61

Problem with RSSI-based distance – not accurate

Pathloss is NOT only due to distance

RSSI is way worse at this than you hope it would be

21

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

22 of 61

Apple uses BLE on its devices to implement “Continuity”

Key part of Apple’s “ecosystem”, including:

Handoff

start tasks on one device and continue on another device

Universal Clipboard,

copying of data from one Apple device and pasting on another

iPhone Cellular Calls

make calls using iPhone’s cellular connection on Mac or iPad

22

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

23 of 61

Uses BLE ”Manufacturer Specific Data” to communicate

Communication with only nearby devices

23

Martin, Jeremy, et al. "Handoff all your privacy–a review of apple’s bluetooth low energy continuity protocol." Proceedings on Privacy Enhancing Technologies 2019.4 (2019): 34-53.

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

24 of 61

Local Communication: Exposure Notifications

Apple and Google collaboration to use phones for contact tracing

Smartphone constantly broadcasts identifier.
Periodically, each smartphone listens for broadcasts around it.
Check list of identifiers to see if you’ve been around someone who is sick.

Requires government/healthcare�system interactions to determine when�an identifier should be flagged as sick

24 states adopted this

Implemented at OS level in background

24

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

25 of 61

Cautionary Tale:�Technology cannot solve all the world’s problems

What was the state of CA Notify on October 4, 2022?

More seriously, advertisement-only was always going to be limited
Bad/imperfect technology is not neutral — this caused measurable harm

Post-hoc analyses show contact-tracing apps reduced trust in government, reduced willingness to use future apps (independent of source)

25

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

26 of 61

Outline

BLE roles

Advertising
Scanning

Advertisement-based Networking?

Advertisement Use Cases
Energy Use
Packet Collisions

26

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

27 of 61

Paper: power measurements of BLE advertisements

Schrader, Raphael, et al. "Advertising power consumption of bluetooth low energy systems." 2016 3rd International Symposium on Wireless Systems within the Conferences on Intelligent Data Acquisition and Advanced Computing Systems (IDAACS-SWS). IEEE, 2016.

27

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

28 of 61

Energy model for BLE advertisements

28

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

29 of 61

Power measured for BLE adv states for hardware

Power use and duration energy measured on

nRF51 Development Kit (nRF51822 System-on-Chip)
nRF52 DK (nRF52832 SoC)

29

nRF51 DK

nRF51822 SoC

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

30 of 61

Exercise: How much energy does one advertising cycle consume?�(for the peripheral / advertising device)

nrf52 chip
0 dBm TX power
30 byte advertisement (total)
Receive window is 100 µs
500 ms advertisement interval
Ignore startup, transition, and processing energy use

30

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

31 of 61

How much energy does one advertising cycle consume?

Energy = TX + RX + sleep
TX:

30 bytes @ 1 Mbps
3 channels
240 us * 3 = 720 us
720 us * 33.2 mW = 23.9 µJ

RX:

RW = 100 us
3 channels
300 us * 32.4 mW = 9.7 µJ

Sleep

500 ms interval
Radio active = 720 us + 300 us
Sleep = 500 ms – 1.02 ms
498.98 ms * 0.0058 mW = 2.9 µJ

Energy = 23.9 µJ + 9.7 µJ + 2.9 µJ = 36.5 µJ

31

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

32 of 61

How much energy does it cost to send data over advertisements?

Configuration

nRF51822 microcontroller
Maximum payload size
+4 dBm transmit power
Connectable advertisement
Sleep power 11 µW

One packet per second example:

110 µW average
~270 days on a CR2032

One packet per minute example:

13 µW average
~2250 days (6 years) on a CR2032

32

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

33 of 61

Outline

BLE roles

Advertising
Scanning

Advertisement-based Networking?

Advertisement Use Cases
Energy Use
Packet Collisions

33

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

34 of 61

Example of real-world research:�Sensor data collection via BLE advertisements

Motivation:

Very resource-constrained device
Diverse environments (“quick-deploy”)

Requirements:

Report data so gateways and users can retrieve it simultaneously
Easy introspection during a deployment

Idea: Ignore difficult questions about networking

Just broadcast the data!
Will it work?

34

DeBruin, Samuel, et al. "Powerblade: A low-profile, true-power, plug-through energy meter." Proceedings of the 13th ACM Conference on Embedded Networked Sensor Systems. 2015.

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

35 of 61

Questions about network capability

What are the odds that a transmitted advertisement will be received?

Packet reception rate

If M redundant advertisements are sent instead, what are the odds that at least one are received?

Data reception rate

How do these odds vary with number of devices, advertising interval, and packet size?

35

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

36 of 61

BLE advertisements are periodic, broadcast transmissions.

Advertisement events occur periodically (T_{adv_interval}: 20 ms–10 s).
Random delay appended before each transmission (t_{adv_delay}: 0–10 ms).
Data payload of up to 31 bytes.

36

time

T_{adv_interval}

+ t_{adv_delay}

advertisement event

Bluetooth SIG. Bluetooth Core Specification 4.2. (2014).

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

37 of 61

What causes transmissions not to be received?

Not within range of the gateway

Or various other losses within the gateway itself�

Two devices try to send at the same time (packet collision)

37

broadcast domain

X

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

38 of 61

What is the probability of a packet collision?

38

Packet 0

time

t_{adv_0}

Jeon, Wha Sook, et al. "Performance analysis of neighbor discovery process in bluetooth low-energy networks." (IEEE Transactions on Vehicular Technology, 2016).� Perez-Diaz de Cerio, David, et al. "Analytical and experimental performance evaluation of BLE neighbor discovery process including non-idealities of real chipsets." (Sensors, 2017).

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

39 of 61

Let’s define some terms for a single advertiser

39

Packet 0

time

T_adv

Jeon, Wha Sook, et al. "Performance analysis of neighbor discovery process in bluetooth low-energy networks." (IEEE Transactions on Vehicular Technology, 2016).� Perez-Diaz de Cerio, David, et al. "Analytical and experimental performance evaluation of BLE neighbor discovery process including non-idealities of real chipsets." (Sensors, 2017).

T_adv: Transmission Time for packet

t_{adv_delay}: Random advertisement delay

T_advInterval: Advertisement interval

Next Packet

T_{adv_interval}

t_{adv_delay}

T_adv

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

40 of 61

What is the probability of a packet collision?

40

Packet 0

Packet 1

time

t_{adv_0}

t_{adv_1}

Packet 1

Jeon, Wha Sook, et al. "Performance analysis of neighbor discovery process in bluetooth low-energy networks." (IEEE Transactions on Vehicular Technology, 2016).� Perez-Diaz de Cerio, David, et al. "Analytical and experimental performance evaluation of BLE neighbor discovery process including non-idealities of real chipsets." (Sensors, 2017).

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

41 of 61

What is the probability of a packet collision?

41

Packet 0

Packet 1

time

t_{adv_0}

t_{adv_1}

Probability of Collision

Jeon, Wha Sook, et al. "Performance analysis of neighbor discovery process in bluetooth low-energy networks." (IEEE Transactions on Vehicular Technology, 2016).� Perez-Diaz de Cerio, David, et al. "Analytical and experimental performance evaluation of BLE neighbor discovery process including non-idealities of real chipsets." (Sensors, 2017).

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

42 of 61

Collisions occur if any part of the two packets overlap

42

Packet 0

Packet 1

time

T_adv

Packet 1

Jeon, Wha Sook, et al. "Performance analysis of neighbor discovery process in bluetooth low-energy networks." (IEEE Transactions on Vehicular Technology, 2016).� Perez-Diaz de Cerio, David, et al. "Analytical and experimental performance evaluation of BLE neighbor discovery process including non-idealities of real chipsets." (Sensors, 2017).

no collision

T_adv

Packet 1

t

t - T_adv

t + T_adv

2 * T_adv

Vulnerable period

no collision

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

43 of 61

Computing probability of collision from vulnerable period and Tx window

43

Packet 0

time

T_adv

t

t - T_adv

t + T_adv

2 * T_adv

Vulnerable period

Packet 0

T_{adv_interval +}t_{adv_delay}

t_{adv_delay}

Transmission window

Probability of Collision

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

44 of 61

Break + Determine Probability of Multiple Failures

Given:

Probability of Collision

Determine:

Probability of Reception for data sent redundantly across M packets

44

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

45 of 61

Easy way to combat inevitable collisions: Just say it twice!�(…or three times, or four times, …)

45

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

46 of 61

Determine Probability of Multiple Failures�(How many times should we say the same thing?)

Given:

Probability of collision of 1 packet with reference packet

Determine:

Probability of Reception for data sent redundantly across M packets
i.e., what are the odds that at least one of the packets doesn’t collide
1 – (Probability of Collision^M)

(P_c)^M = Probability that all of them collide
1-that = Probability that NOT all of them collide

46

Probability of Collision

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

47 of 61

When are transmissions from two devices independent?

Assumption is true for any BLE device that has been advertising for some time

Sum of random delays grows the uncertainty of transmission.
Applied to periodic transmissions, any point in interval becomes equally likely.

Range of 1x delay, 2x delay, 3x delay, until it wraps

47

time

T_{adv_interval}

delay

T_{adv_interval}

delay

T_{adv_interval}

delay

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

48 of 61

When are transmissions from two devices NOT independent?

Independence assumption is false for two BLE devices that have recently collided.

If T_{adv_interval} is identical, next transmissions with be close in time.
Collision is determined by difference of random delays.
Further repeat collisions have the same probability of occurrence.

48

time

T_{adv_interval}

delay

New time period collisions could occur within

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

49 of 61

Calculating probability of a repeat collision

Difference of two uniform random delays forms a triangular distribution

49

difference in transmission time

probability

t_{adv_0}

t_{adv_1}

Probability of Repeat Collision

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

50 of 61

Important lesson: spend time on things that are important

How important is accounting for repeat collisions with advertisement-based networking?�

Maximum error is about a 1% change in Data Reception Rate.

This is due to size of delay 10 ms compared to size of transmission ~300 µs.

50

200 ms interval advertisements

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

51 of 61

Equations for modeling data transmissions

51

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

52 of 61

Is the model valid?

Empirical testing setup:

50 devices
1 meter from scanner
5-10 cm apart�

Transmit monotonically increasing sequence numbers.�

Sweep number of devices and advertising intervals.

52

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

53 of 61

The model is accurate across advertisement rates and deployment sizes.

Accuracy is fairly consistent across intervals.

The model consistently overestimates the measured PRR values.

The effect could be due to RF interference.

53

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

54 of 61

The model accurately accounts for redundancy as well.

The same dataset can be used to measure the effect of redundancy by grouping sets of sequence numbers.

The model again slightly overestimates, but error reduces quickly as DRR approaches 100%.

54

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

55 of 61

What questions can we answer with a collision model?

Original questions

What are the odds that a transmitted advertisement will be received?
If M redundant advertisements are sent instead, what are the odds that at least one are received?
How do these odds vary with number of devices, advertising interval, and packet size?

Additional questions

Can redundancy make advertisements reliable?
Is it better to transmit often for high redundancy or rarely for less congestion?

55

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

56 of 61

Redundancy results in high DRR even with many devices.

In this example, a sensor has new data once per second and sends it in 1-3 packets.

Even without redundancy, data reception rates never fall below 87% even with 200 devices in a deployment, assuming no interference.

56

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

57 of 61

Redundancy is (normally) better than less congestion.

Design question:

Send more packets to gain from redundancy?��OR�
Send less packets to reduce congestion?

The answer changes, but only with many devices.

57

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

58 of 61

Outline

BLE roles

Advertising
Scanning

Advertisement-based Networking?

Advertisement Use Cases
Energy Use
Packet Collisions
What about Scan Requests & Responses?

58

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

59 of 61

Scan requests/responses seem intriguing

Why not send most data in scan responses instead of advertisements?

Theoretically could reduce energy costs

Can we use scan requests as a form of acknowledgement?

Could relieve need for redundant transmissions

Problem: scan requests/responses don’t work all that well

59

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

60 of 61

Scan Requests and Responses don’t work for data use case

Goal: provide a little extra advertisement data on demand

Problem: exponential backoff for lost messages

If there is a request without a response, scanners assume collision with another scanner and exponentially back off from requesting
But collisions are far more likely between a device and a scanner, which should not have back off
Result is that scan requests will occur far less frequently than expected
Instead, better just send additional advertisements with different data

Kravets, Robin, Albert F. Harris III, and Roy Want. "Beacon trains: blazing a trail through dense BLE environments." Proceedings of the Eleventh ACM Workshop on Challenged Networks. 2016.

60

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell

61 of 61

Next up: BLE Connections

61

CSE 122/222C ; WES 269 [WI25]

CC BY-NC-ND Pat Pannuto – Content developed in coordination Branden Ghena and Brad Campbell