1 of 110

CSE 122 / 222C ; WES 269�Wide Area Communications and the Evolution of Cellular

Pat Pannuto, UC San Diego

ppannuto@ucsd.edu

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

2 of 110

Cell Goals

Introduction to Wide Area Communication

Understand fundamentals of cellular technology

Learn how “older” cellular technology works

And why it is relevant to the IoT today

Understand emerging, “IoT-friendly” updates to cell

Pros, cons, and pitfalls of using cellular networks for IoT deployments

2

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

3 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

3

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

4 of 110

Wide area networks

Communication at the region/city scale rather than the building/residence scale

Throughout cities
Agricultural deployments
Industrial facilities

City-scale sensing is one very popular domain

What might we want to sense throughout a city?

4

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

5 of 110

Air quality monitoring

5

[1] Cheng et al. AirCloud: a cloud-based air-quality monitoring system for everyone. 2014. [2] Purple Air. 2018.

[1]

[2]

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

6 of 110

Audio detection, classification, and localization

6

[1] Girod et al. The Design and Implementation of a Self-Calibrating Distributed Acoustic Sensing Platform. 2006. [2] Lédeczi et al. Multiple Simultaneous Acoustic Source Localization in Urban Terrain. 2005. [3] Sounds of New York City. 2016.

[1]

[2]

[3]

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

7 of 110

Traffic queue sensing and congestion control

7

[1] Sen et al. Kyun Queue: A Sensor Network System To Monitor Road Traffic Queues, 2012

[1]

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

8 of 110

Array of Things project - Chicago

8

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

9 of 110

9

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

10 of 110

Considering wide-area networks

10

Data

Throughput

Range

WiFi

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

11 of 110

Considering wide-area networks

11

Bluetooth

WiFi

BLE

Zigbee

Data

Throughput

Range

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

12 of 110

Considering wide-area networks

12

Bluetooth

WiFi

BLE

Zigbee

Data

Throughput

Range

Lower Power�& Lower Cost

Higher Power�& Higher Cost

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

13 of 110

Considering wide-area networks

13

Bluetooth

WiFi

2G

3G

4G

BLE

Zigbee

Data

Throughput

Range

Lower Power�& Lower Cost

Higher Power�& Higher Cost

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

14 of 110

Takeaway: Cellular provides the IoT the only reliable global coverage available today

If the goal is deploy-today + work-anywhere, cell is the only option

(Or arguably satellite)

That’s not to say cell works everywhere!

Just the best-available

14

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

15 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

15

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

16 of 110

Poll: What decade was the first commercial mobile phone?

1910-1920
1920-1930
1930-1940
1940-1950
1950-1960
1960-1970
1970-1980
1980-1990
1990-2000

16

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

17 of 110

The first commercial mobile phone was in 1946!

By 1948, 5000 users placing around 30,000 calls weekly

17

https://www.smithsonianmag.com/innovation/first-mobile-phone-call-was-made-75-years-ago-180978003/

“The equipment, of course employing vacuum tubes, weighed eighty pounds, filled much of a vehicle’s trunk and drew so much power that it would cause the headlights to dim. Service cost $15 per month, plus thirty to forty cents per local call, equivalent to $175 2012 dollars, plus $3.50 to $4.65 per call. “

https://ethw.org/The_Foundations_of_Mobile_and_Cellular_Telephony

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

18 of 110

The first `mobile` network architecture

Single transmitter, multiple receiver

Why?

Medium Access Control?

Push-to-talk [half-duplex]
City-wide “party line”
Maximum of three concurrent callers

18

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

19 of 110

1965: The Improved Mobile Telephone Service

”More spectrum”

Bell: VHF Low (35-44 MHz, 9 channels), VHF High (152-158 MHz, 11 channels), and UHF (454-460 MHz, 12 channels) — It’s good to be a monopoly [well 85%]
7 channels at VHF, and 12 channels at UHF for "RCCs" (Radio Common Carriers)

These will end up becoming pagers, mostly

“Full Duplex”

Implemented as channel pairs (two half-duplex)

Spectrum alone is not enough

Limited to 40,000 total subscribers
E.g. NYC had 2,000 people on 12 channels with a typical wait of 30 minutes to place a call

19

Phones fit in a briefcase now

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

20 of 110

Quick review: How can we share spectrum?

Frequency division

Already doing that

Time division

Already doing that

Code division

Won’t come to [US…] mobile networks until the 80’s

…?

20

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

21 of 110

What’s wrong inefficient with this picture?

Early mobile networks handled the mobility challenge with wide-area coverage
The active user could be anywhere in the 40~60 mile range of the transmit tower

So they had exclusive access to that whole area!

12 frequency channels 🡪 12 users

21

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

22 of 110

The `cell` in cellular networks

Cell is spatial division

Reduce transmit power, smaller area

Now…

12 / 3 = 4 channels per cell
* 6 cells in area
= 24 users of same spectrum!

Choose cell size (aka, towers deployed $$) based on needs

22

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

23 of 110

Spatial division tradeoffs

Need more infrastructure (more towers)
Need logic to support handoff

Cellular concept in Bell labs at 1947 [one year after mobile], but no handoff

23

?

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

24 of 110

How often does handoff need to happen anyway?

Really a question of `how big are cells`
For 1G, a back-of-envelope answer:

24

	Speed (mph)	Speed (kph)	Cell Radius (km)	Average # of Handoffs for a 3-minute call
Freeway, rural	65	104	16	0.33
Freeway, urban	65	104	1.6	3.25
Surface streets, urban	30	48	1	2.4
Pedestrian, urban	1.5	2.4	1	0.12
Pedestrian, microcell	1.5	2.4	0.1	1.2

Table data from Arthur H. M. Ross, PhD: http://www.cdg.org/technology/cdma_technology/a_ross/handoff.asp

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

25 of 110

Introducing 1G!

AMPS — Advanced Mobile Phone Service (circa 1983)

Other 1G’s:

Nordic Mobile Telephone (NMT) — Nordics, Eastern Europe, Russia
Total Access Communications System (TACS) — UK, West Germany, Portugal South Africa
Radiocom 2000 — France
…

1G is defined by:

Analog voice channels
Established by digital control channel 🡨 handoffs

25

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

26 of 110

Quick Aside: DTMF Signalling

How do you dial a phone number?

Early on: switchboard operators [physically plug wires to make connections]
Pulse dialing:

Dual Tone Multi Frequency:

What info for handoff?

26

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

27 of 110

How to implement handoff?

<brainstorm>

27

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

28 of 110

How does AMPS implement handoff?

Base station detects weak signal
Base station tells mobile about new, better tower
Base station sends cutover trigger

Lots bad about this:

No `make before break` — many drops!
Mobile has better estimates than base station

28

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

29 of 110

Spatial division in practice

Directional antennas at tessellation corners provide redundant coverage [robustness!]

29

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

30 of 110

Spatial division in practice

Directional isolation is imperfect, requires many bands in practice

30

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

31 of 110

Over time, AMPS did also get more frequencies

AMPS in the US:

850 MHz band
Divided into “A” block and “B” block

Granted to local phone company and ‘wireless company’ respectively

Each block had 21 control, 395 voice channels

Several expansions of this frequency block

Most interesting, in 1989, UHF TV channels 70-83 added 666 more channels

31

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

32 of 110

Cellular phone or walkie-talkie?�https://www.youtube.com/watch?v=JsEgUA4z_SM

32

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

33 of 110

The first “sunset”

Single-user analog channels is not very efficient
Wireless carriers want to replace 1G

Key Takeaway: Spectrum allocation is the most scarce resource, therefore want to maximize efficiency of its use

efficiency, n., paying customers / spectrum?

But many things in the built environment rely on AMPS

Because when they were built, AMPS was the most reliable

OnStar
ADT home alarms [got sued!]

What options do you have as the owner of an AMPS device during a sunset?

33

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

34 of 110

Poll: When did AMPS turn off in the United States?�(turned on ~1979)

1980-1985
1985-1990
1990-1995
1995-2000
2000-2005
2005-2010
2010-2015
2015-2020
AMPS is still active

34

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

35 of 110

Aside: Super cool hackaday projects

Building your own 1G network!

https://hackaday.com/tag/amps/
https://github.com/unsynchronized/gr-amps
⚠ ⚠ ⚠ This uses spectrum you do not own ⚠ ⚠ ⚠

“The FCC can levy fines of up to $11,000, per transmitter, per day, with up to six months in federal prison, for violations of the Code of Federal Regulations.”
Could deploy with wired connections though…

35

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

36 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

36

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

37 of 110

IS-54 aka 2G aka TDMA aka D-AMPS (Digital AMPS)

Deployed ~1993
Used TDMA to share AMPS channels

Each AMPS analog channel-pair was 30 kHz
Digital compression allowed for 3 time slots per channel — tripled capacity!

Later iterations would compress further, 6 time slots per channel
Without additional bandwidth: 6*56 = 336 simultaneous devices per cell

Beat out competing FDMA analog solution from Motorola

Narrowband AMPS, “N-AMPS”, compressed calls to fit in 10 kHz channels

Also encrypted [poorly: CMEA broken publicly in 1997]

37

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

38 of 110

GSM aka 2G aka TDMA

Wait, didn’t the last slide say 2G was something else?

What is “2G” then?

Exactly.
“2G” isn’t one technology, it’s a collection of “second generation” protocols
This is true for later generations too

38

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

39 of 110

The first 2G sunset happened in 2008

D-AMPS was sunset when AMPS was sunset
Modern “2G” is all GSM
And if you think this is confusing, it only gets worse

Anecdata: Pat as a Qualcomm intern in 2010 asked everyone I met there “what actually is 4G,” and every engineer just laughed at me

39

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

40 of 110

GSM

So can we just call modern 2G “GSM”?
“Global System for Mobile Communications”

It is a standard [the dominant 2G]
GSM meant “phone” for a while in parts of Europe

(and apparently still in Belgium?)

GSMA is the GSM Association; which a lot of people just call GSM

Because GSM made [a] 3G and LTE [which is a 4g?] and 5g [ish! more next time]

Yay.

40

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

41 of 110

Origins of GSM

An incredible sweet spot of timing, technology need, and politics?

1987: 13 European countries sign accord to use one wireless standard

10 months later: GSM standard ratified
Insight: Not a ton new in GSM by ‘87, just a lot of arbitrary design decisions

41

Not a very `mobile` mobile network…

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

42 of 110

The real innovation of GSM was the indirection and interoperability it enabled

Europe hit this problem before the US because the US is large

(and US telecoms owned/operated subsidiaries in Canada)

Coincided with major growth in global mobility

Mobile devices needed to be able to use other carrier’s infrastructure

Early GSM isn’t anything fancy in modulation, data, etc.

It’s all about the control channels

42

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

43 of 110

43

https://en.wikipedia.org/wiki/File:Gsm_structures.svg

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

44 of 110

Circuit switching

44

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

45 of 110

Comparing circuits vs packets

Circuit switching

Virtual electrical connections between two devices
Full capacity is dedicated to only those two devices

Packet switching

Chunks of data get routed between devices
Overhead to figure out where they go

45

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

46 of 110

GSM evolved over time to add data

GSM was voice-only at first
Then overlaid data on the voice channel

Circuit Switched Data (CSD)

9.6 — 14.4 kb/s
Previous hacks were literally cell phone calling a modem (~3.5 kb/s); same concept

Then added actual packetized data [“2.5G”]

General Packet Radio Service (GPRS)

Throughput 9.2 – 21.55 kbit/s/timeslot [goodput 8-20 kbit/s/timeslot]

Enhanced Data Rates for GSM Evolution (EDGE)

Why is packetized more efficient use of spectrum than circuit switched?

46

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

47 of 110

So why does the IoT care about 2G?

It’s easy

2G == GMSK, 3G = QPSK [requires linear PA], 4G = …

It’s cheap

Relative to other cellular plans, anyways

It’s low power*

*For select IoT workloads
Energy-per-bit of continuous stream improves each generation
Energy-per-event of infrequent events is trickier though

47

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

48 of 110

Why might 2G continue to survive [globally]?

It’s cheaper coverage

2G has longer range than 3G

It fills the legacy niche

High performance HW includes fallbacks

2G/3G radios, then 2G/3G/4G radios, now 2G/3G/4G/5G radios
Consider the iPhone 13 Pro:

Brick phones still popular!

48

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

49 of 110

2G in the US…�[Plans…]

AT&T stopped servicing 2G networks in 2016
Verizon Wireless phased out its 2G CDMA network at the end of 2020
Sprint will shut down their 2G CDMA network in April 2022

Unclear if this happened or not

T-Mobile plans on 2023�April 2024�No firm date at this time

49

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

50 of 110

Does GM ever learn?�[Could they have?]

What would you put in cars if you were the PM of OnStar today?

50

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

51 of 110

Break + xkcd

51

https://xkcd.com/925/

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

52 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

52

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

53 of 110

Starting with the end: the 3G sunset

[As of Oct, 2021] announced retirements:

Verizon: Dec 2022
AT&T: Feb 2022
T-Mobile:

For Sprint customers / bands: Jan 2022

T-Mobile delays shutdown of Sprint 3G Network [now Mar 2022]
Anecdata: The phone that started this presentation was a Sprint phone that needed to be replaced; went to T-Mobile in mid-Oct; got new phone on Dec 30 #ChipShortagesAreHardYo

For original T-Mobile: July 2022
When do you think T-Mobile’s 2G sunset is scheduled for?

Last scheduled Dec 2022… I’ll believe that when I see it
n.b. Verizon and AT&T actually did sunset 2G

53

AT&T: 2.7% of customers will be affected by 3G shutdown

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

54 of 110

Checking in on T-Mobile?

https://www.t-mobile.com/support/coverage/t-mobile-network

(Accessed: Jan 20, 2023)

54

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

55 of 110

Checking in on T-Mobile?

https://www.t-mobile.com/support/coverage/t-mobile-network

(Accessed: Feb 10, 2025)

55

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

56 of 110

Why does 2G have staying power that 3G doesn’t?

T-Mobile’s letter to Congress on 2g/3g sunset

n.b. Senator Schatz seems to get this stuff

56

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

57 of 110

So what did [*]G provide over 2G anyway?

Biggest thing is throughput
And a spectrum land-grab

Wireless providers advocated [successfully] for [global] release of additional spectrum to support new, faster wireless
Lots of [initial] 4G rollout also on new spectrum

But no more coming — sunsets/whitespaces only way to get more 4G spectrum

5G is [mostly] on much higher bands, not suited for 4G

57

Terahertz Band: The Last Piece of RF Spectrum Puzzle for Communication Systems (preprint; Elayan et al.)

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

58 of 110

Aside: Eventually, the naming thing got annoying, so an international standards body fixed* it�*Yeah, you know how this story will end

The International Telecommunications Union (ITU) is a UN branch

ITU-R set standards for what can be considered “a 4G technology”

”Generations” are now defined by speed

E.g. to qualify as “a 4G technology”: 100 Mb/s+ (mobile), 1 Gb/s+ (stationary)
Fun game: How fast is LTE?

LTE spec maxes out at 300 MB/s down and 75 MB/s up
It seems LTE has been retroactively classified as 3.95G

[By everyone except the wireless carriers?]

58

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

59 of 110

3GPP�aka: the actual answer for what stuff is really doing

3^rd Generation Partnership Project (3GPP)

Industry alliance for development of telecoms standards

Established around 1998
Makes “Releases” which are roughly analogous to IEEE standards/versions

Release 8 (2008) LTE ~4G
Release 15 (2018) NR (New Radio) ~5G

Focused on the practical

ITU post-hoc defined “4G”, 3GPP defined LTE and LTE_________

59

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

60 of 110

Mapping ”4G”, “LTE”, “LTE Advanced”, etc onto actual technologies

60

This Qualcomm presentation is great: https://www.qualcomm.com/media/documents/files/demystifying-3gpp-and-the-essential-role-of-qualcomm-in-leading-the-expansion-of-the-mobile-ecosystem.pdf

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

61 of 110

LTE Categories

Different equipment supports different “categories” of LTE

Maximum MCS index supported

Examples

iPhone 6 (2015): Cat 4
Pixel 3 (2018): Cat 16

Aside: Hey look, some LTE is ”ITU 4G”!

61

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

62 of 110

Corollary to improved throughput is improved energy performance

Same philosophy as high-performance architecture, etc

There is always baseline load, which going faster amortizes

62

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

63 of 110

But energy per transaction is complex

Radios, cellular protocols, and cellular providers have state machines

Hard to predict a priori and globally ☹

63

Characterizing radio resource allocation for 3G networks; Feng Qian, Zhaoguang Wang, Alexandre Gerber, Zhuoqing Morley Mao, Subhabrata Sen, Oliver Spatscheck; SIGCOMM’10

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

64 of 110

Numbers from one implementation

Take these as a reference point [from 2013!], not gospel

64

The System Power Control Unit Based on the On-Chip Wireless Communication System

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

65 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

65

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

66 of 110

So what is 5G doing differently?

Largely, 5G is a small(er)-cell technology

“Urban optimized”
Unlikely to roll out to all geographic areas ever (because of infrastructure cost)

This does also allow for higher performance

66

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

67 of 110

Aside: One last bit on emerging new G’s�aka: What was up with the 5G / airplane kerfuffle?

aka: when FDMA doesn’t do the “D” part so great

n.b. The FAA site on this is�actually pretty good, and�both accessible / technical

https://www.faa.gov/5g

67

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

68 of 110

Filter design is hard ($$), a touch on the EE black magic side (especially equations vs practice)

Graphics stolen from https://www.5gtechnologyworld.com/5g-altimeter-interference-aviation-versus-telecoms/ , the FAA site, and a quick google image for ‘higher order filter cutoff’

68

160 mph (~landing speed)

*96s ~= 4.25 mi

*20s ~= 0.9 mi

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

69 of 110

Resource: Some of the best technical details I could find

National Business Aviation Association (NBAA) webinar

https://www.bigmarker.com/nbaa/NBAA-News-Hour-What-the-Looming-Threat-From-5G-Interference-Could-Mean-for-Your-Flight?show_live_page=true&add_to_calendar=true&bmid=859a33fdec7a
One takeaway from webinar: Airports aren’t the problem; helos, emergency, etc

69

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

70 of 110

Highlights from the webinar [n.b. recorded Dec 7, 2021]

70

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

71 of 110

Just because I was curious…

n.b. “SoCal TRACON” is the busiest ATC in the world [SD based; incl LA]

71

160 mph (~landing speed)

*96s ~= 4.25 mi

*20s ~= 0.9 mi

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

72 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

72

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

73 of 110

Wait, doesn’t this class focus on “low-power wireless”?

Bad jokes in computer science:

How do you make a supercomputer draw less power?

Turn it off.

73

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

74 of 110

A purely hypothetical motivating example

Research question: What happens to ”recycled” mobile phones?
Research opportunity:

My partner is “trading in” her iPhone6 for an iPhone13
Where does that phone go? What does T-Mobile do with it?
Any guesses?

How might we answer this question?

74

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

75 of 110

These things aren’t creepy at all

It’s actually sad and scary the wealth of ‘covert tracking’ tech

75

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

76 of 110

Limited battery life will save us!

Not really…

76

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

77 of 110

Limited GPS coverage will save us!

Not really…

77

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

78 of 110

Surely people will notice a big, bulky tracker right?

I mean.. I’m hoping not

78

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

79 of 110

For a little bit of time, $$$ might save us

Traditional cellular billing is fixed per-unit not per-data

Given what you’ve learned of cellular infrastructure, why might this be?

But as we will see later, big effort to roll out IoT-aware infra

79

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

80 of 110

Takeaway: Cellular provides the IoT the only reliable global coverage available today (changing fast!)

If the goal is deploy-today + work-anywhere, cell is the only option

(Or arguably satellite)

That’s not to say cell actually works everywhere!

Just the best-available

Contrast: What is the insight behind AirTags?

Or Tile, Cube, etc.?

80

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

81 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

81

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

TODO someday maybe; figure out how to weave the parallel Soviet development path, e.g. from the CDMA History on Wikipedia:

The technology of code-division multiple access channels has long been known. In the Soviet Union (USSR), the first work devoted to this subject was published in 1935 by Dmitry Ageev.^[4] It was shown that through the use of linear methods, there are three types of signal separation: frequency, time and compensatory.^[^{clarification needed}^] The technology of CDMA was used in 1957, when the young military radio engineer Leonid Kupriyanovich in Moscow made an experimental model of a wearable automatic mobile phone, called LK-1 by him, with a base station.^[5] LK-1 has a weight of 3 kg, 20–30 km operating distance, and 20–30 hours of battery life.^[6]^[7] The base station, as described by the author, could serve several customers. In 1958, Kupriyanovich made the new experimental "pocket" model of mobile phone. This phone weighed 0.5 kg. To serve more customers, Kupriyanovich proposed the device, which he called "correlator."^[8]^[9] In 1958, the USSR also started the development of the "Altai" national civil mobile phone service for cars, based on the Soviet MRT-1327 standard. The phone system weighed 11 kg (24 lb). It was placed in the trunk of the vehicles of high-ranking officials and used a standard handset in the passenger compartment. The main developers of the Altai system were VNIIS (Voronezh Science Research Institute of Communications) and GSPI (State Specialized Project Institute). In 1963 this service started in Moscow, and in 1970 Altai service was used in 30 USSR cities.

82 of 110

Sizing networks is hard

Sometimes you just lose performance…

82

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

83 of 110

Traffic fluctuates in all networks, for cellular it becomes a function of both when and where you are

Understanding Mobile Traffic Patterns of Large Scale Cellular Towers in Urban Environment; Xu, Li, Wang, Zhang, Jin; IEEE ToN

83

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

84 of 110

Why does this matter for IoT Deployments?

Say you’re building a power grid sensor you want to deploy globally…

84

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

85 of 110

What is an MVNO?�Mobile Virtual Network Operator

Decoupling the builders of infrastructure from the sellers
Not a new concept, but seeing aggressive growth

MetroPCS; Cricket; Boost Mobile; etc
GoogleFi; Xfinity Cellular

85

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

86 of 110

So how does Particle, a small IoT platform startup, provide global cellular coverage?

We deployed some Electrons in Accra, Ghana sending a message once per minute 24/7 for a few weeks

PRR [POST Reception Rate]?

1: 10% 2: 20% 3: 30% 4: 40% 5: 50% 6: 60% 7: 70% 8: 80% 9: 90%

Almost zero PRR from ~7-9am and ~4-7pm daily

Introducing traffic priority

Call Particle: “What gives?”

Particle buys from T-Mobile
T-Mobile buys from Deutsche-Telekom [didn’t own them yet]
DT buys from Vodafone
Vodafone buys from MTN
MTN has 5 tiers of traffic priority on their network [guess which tier we were in?]

86

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

87 of 110

So how do you get higher on the priority list?

You buy from MTN
… also not easy

Limit of 3 SIMs / person due to fraud

Particularly important due to prevalence of SIM-based mobile money

87

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

88 of 110

So how do you deploy in Tanzania?

Not limited in SIMs, but limited in payment plans
Post-paid plans not an option

Need to purchase ‘airtime recharges’
Which you use by texting from that phone to an SMS shortcode

So now you must have in-country staff!

88

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

89 of 110

So how do you actually realize this claim?

Good. Question.

89

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

90 of 110

On your own: Understanding global cellular

No lab on cell
Instead: Research assignment

What does cellular look like around the world?

90

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

91 of 110

Outline

Wide-Area Network Background
Evolution of Cellular Network Technologies

1G
2G
3G, 4G
5G and beyond

Cell and IoT in Practice

Cell’s Wide Area Moat
Understanding MVNOs

Emerging Technologies: NB-IoT, LTE-M

91

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

92 of 110

Additional low-end categories for IoT

LTE Cat 0

Traditional LTE, but focused on the really low end

LTE-M (LTE Cat M1)

375 kbps uplink, 300 kbps downlink (for the actually implemented mode)
Reduced power and maximum bandwidth
Increased range

NB-IoT (LTE Cat NB1)

65 kbps uplink, 26 kbps downlink
Reduced power and greatly reduced bandwidth
Greatly increased range

92

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

93 of 110

Why do we need “special categories” for IoT on cell?

We can treat IoT devices differently than human-centric devices
Pragmatic for the end device

Lower power
Allow for long-off periods

Pragmatic for network operators

Allows for scale– network no longer needs to assume that devices could always be on in each cell

93

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

94 of 110

LTE-M and NB-IoT were developed in parallel

94

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

95 of 110

LTE-M and NB-IoT downlink and uplink

OFDMA downlink

Put the more complicated hardware in the cell tower [simple FFT demodulator]

SC-FDMA (single carrier FDMA) uplink

Blocks of subchannels combined into one signal
Similar concept, but simpler for end devices to implement

95

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

96 of 110

LTE resource allocation

Cellular uses OFDMA to schedule

Time + Frequency -> “2D Scheduling”

Cellular uses single channels up to 20 MHz

Further divides these into 100 Resource Blocks

Resource Block

12 subcarriers for OFDM in frequency (15 kHz each)
7 symbols in time (0.5 ms)

Devices are allocated frequency and time based on what they are sending

Allocated in units of Resource Blocks

96

One “Resource Block”

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

97 of 110

Resources used by LTE-M and NB-IoT

LTE-M uses up to 6 resource blocks

1.4 MHz of bandwidth (1.080 MHz)
Can co-exist with other normal LTE traffic, scheduled by cell tower
Limited to only some capability of LTE

97

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

98 of 110

Resources used by LTE-M and NB-IoT

NB-IoT uses up to 1 resource block

200 kHz of bandwidth (180 kHz)
Multiple deployment options

Guard-band in practice

98

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

99 of 110

Reducing energy use for IoT devices

Reduce max Tx power to 20 dBm

Increased receive sensitivity at tower will cover it

Extended Discontinuous Reception (eDRX)

Allow devices to reduce paging period and still stay on network
Cell tower will hold messages

What does this get to?

“For a LTE-M1 device that transmits data once per day, and wakes up every 60 hyper frames to check for commands (this would be about every 10 minutes), a life of 4.7 years is achievable on 2 AA batteries.”

99

Graphics, quote from https://www.link-labs.com/blog/lte-e-drx-psm-explained-for-lte-m1

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

100 of 110

Further power reduction for simple devices

Power Saving Mode (PSM)

For very simple, uplink-focused devices, allow them to turn off entirely but stay connected�
Minutes to days in duration

Notify tower before sleeping, listen for packets after each transmission

100

Graphics from https://www.link-labs.com/blog/lte-e-drx-psm-explained-for-lte-m1

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

101 of 110

Some numbers from an actual telecom: Aeris�[n.b. Aeris has been a leader in cellular M2M since the 90’s]

PSM has two timers, devices request values, tower chooses actual:

Extended Timer (“sleep” timer)

3GPP max is 35,712,00s [413.33 days]
Aeris timer range: Min 240m [4h]; Max 413 days
“Aeris Fusion” timer range: Max: 12.9 days

Active Timer (how long will the device stay in idle after communication?)

101

Numbers from https://aeriscom.zendesk.com/hc/en-us/articles/360049848254-Understanding-LTE-M-Power-Management-Modes

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

102 of 110

Improved range for LTE-M and NB-IoT

LTE defines a Maximum Coupling Loss (MCL) a.k.a Link Budget

Traditional cellular: 144 dB (~2.5 km)
LTE-M: 160 dB (~5 km)
NB-IoT: 164 dB (~10 km)

Sigfox: ~155 dB
LoRaWAN: ~143 dB

Note that many cellular bands are often on higher frequencies

Example: 1900 GHz
Coarsely, lower frequency is longer range, but it’s complicated

102

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

103 of 110

Coarsely, lower frequency -> longer range

This was the picture circa 2019
Why else might T-Mobile have really wanted to buy Sprint…

103

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

104 of 110

Cellular deployments

Originally unclear which would be dominant

Verizon and AT&T focused on LTE-M
T-Mobile focused on NB-IoT
All rolled out services nationwide in the 2018-2019 timeframe

Networks are expanding to provide both capabilities

LTE-M: AT&T, T-Mobile, Verizon, US Cellular
NB-IoT: AT&T, T-Mobile, Verizon

Pricing models still very uncertain

NB-IoT example: $5 per device per year up to 12 MB, 10 packets per hour
Future adoption will greatly depend on these

104

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

105 of 110

105

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

106 of 110

Microcontroller support

Devices need to be certified

Hardware and software
Tend to be modules or dual-core systems

Add a SIM card to connect to network

106

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

107 of 110

What about 5G?

NB-IoT and LTE-M are the low-power, wide-area 5G solutions

Intent is to coexist with 5G solutions for human-centric devices

Even if 4G sunset occurs, LTE-M and NB-IoT will still be around

107

https://www.ericsson.com/en/reports-and-papers/nb-iot-and-lte-m-in-the-context-of-5g-industry-white-paper

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

108 of 110

Break + Open Question

Cellular hardware almost always requires certified radio modules where you can’t change the code at all. Why?

108

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

109 of 110

Break + Open Question

Cellular hardware almost always requires certified radio modules where you can’t change the code at all. Why?

Otherwise you could cheat at the protocols!!
Or just generally not follow them fairly.

Avoids “tragedy of the commons” by allowing specific trusted devices only

109

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]

110 of 110

Up Next: LPWANs (in ISM bands)

110

CC BY-NC-ND Pat Pannuto – Content derived from materials from Branden Ghena and Brad Campbell

CSE 122/222C ; WES 269 [WI25]