How do telephones work?

If I can expand a bit on the old way of doing things, because I think it's some of the coolest electrical equipment ever that you could actually watch working and understand.

Every subscriber line has its own uniselector. It's a stepping switch with 25 positions, each position connected to a first selector. When you pick up the phone the uniselector steps round its contacts looking for an unused first selector. When it finds one it stops stepping and the subscriber hears a dial tone.

The first selector is a Strowger two-motion selector. It's like a three dimensional uniselector. It has ten banks of ten contacts each and two ratchet-pawl mechanisms, one of which steps a wiper UP to select a certain bank, and the other to step the wiper ROUND the bank. It's hard to find a decent clear photo of one of these, so a drawing will have to do.
When the subscriber dials the first number, the dial pulses make the selector step UP the bank of contacts. Then when the pulses are finished, the selector runs ROUND the bank looking for a free second selector, exactly the way the uniselector did. Each contact on the 2-motion selector is wired to another selector – the first bank leads to selectors for all numbers starting with '1', the second to selectors for all numbers starting with '2', and so on. The final selector in the chain connects not to more selectors, but to 100 individual subscriber lines ending in 00-99. The last-but-one number dialed steps it up to the appropriate row, and the final number steps it round to the right line. The called number then rings and the connection is complete.

When the caller hangs up, their line relay detects it and drops out, and the whole chain of selectors used for that call are released and reset by springs ready for the next call.

What are the differences between 1G, 2G, 3G, 4G and 5G?

What exactly is a 'G' or 'Generation'?

  • In a nutshell, each Generation is defined as a set of telephone network standards, which detail the technological implementation of a particular mobile phone system.

1G – Analog

  • Introduced in 1987 by Telecom (known today as Telstra), Australia received its first cellular mobile phone network utilising a 1G analog system. The analog network was responsible for those bulky handheld 'bricks' that you might have had the displeasure of using and your wallet the displeasure of buying (originally retailed at around $4250).
  • The technology behind 1G was the AMPS (Advanced Mobile Phone System) network. Permanently switched off at the end of 1999, AMPS was a voice-only network operating on the 800MHz band. Being a primitive radio technology, AMPS operated in the same manner as a regular radio transmission, much like your UHF radio where the 800MHz band was split up into a number of channels (395 voice, 21 control) via FDMA (Frequency Division Multiple Access).
  • Each channel was 30KHz wide and could support only one user at any time, meaning that the maximum number of mobile phone users per cell tower was 395. The tower assessed the signal strength of each user and assigned channels dynamically, ensuring that channels could be reused by multiple towers without interference.

Problematic? Yes, and not just a limited number of users.

  • Just like your UHF radio, anyone with a radio scanner capable of receiving/transmitting on the 800MHz band could drop in on your call. Being analog, the 800MHz band was also susceptible to background noise and static caused by nearby electronic devices. However the simplicity of the AMPS design meant it did have one advantage over later 2G networks – coverage. An AMPS user could connect to a cell tower as far as the signal could be transmitted (often >40km depending on terrain).

At its peak, the 1G network had around 2 million subscribers.

2G – Digital

  • Fast forward to 1993 Telecom, now known as Telstra, introduces the digital network. The introduction came about to overcome many of the issues with the AMPS network highlighted above, with network congestion and security being the most important two motivators. With this new technology came many of the services we now take for granted – text messaging, multimedia messaging, internet access, etc, and also introduced us to the SIM card.
  • This fancy new digital network is called GSM – Global System for Mobile Communication, and its technological backbone of choice is TDMA (similar to FDMA). The radio frequency band utilised by GSM is the 900MHz spectrum and later introduced on the 1800MHz band.

So how is this network any better than AMPS?

  • The secret lies in TDMA – Time Division Multiple Access. The FDMA component splits the 900MHz (actually 890MHz to 915MHz) band into 124 channels that are 200KHz wide. The 'time' component then comes into play in which each channel is split into eight 0.577us bursts,significantly increasing the maximum number of users at any one time. We don't hear a 'stuttering' of a persons voice thanks to the wonders of digital compression codecs, which we're not going to go into here.

Aside from more users per cell tower, the digital network offers many other important features:

  1. Digital encryption (64bit A5/1 stream cipher)
  2. Packet data (used for MMS/Internet access)
  3. SMS text messaging
  4. Caller ID and other similar network features.

Problems? You bet.

  • Unlike its AMPS predecessor, GSM is limited severely in range. The TDMA technology behind the 2G network means that if a mobile phone cannot respond within its given timeslot (0.577us bursts) the phone tower will drop you and begin handling another call. Aside from this, packet data transmission rates on GSM are extremely slow, and if you're on Vodafone/3/Virgin/Optus you've probably had first hand experience on this when you go outside your networks defined 'coverage zone'.

To overcome these two problems we're going to introduce two new networks – CDMA and EDGE.

CDMA

  • Code Division Multiple Access. This branch of 2G was introduced by Telstra in September 1999 as a replacement for customers who could receive a good signal on AMPS, but were outside GSM's limited range. The extended range is achieved by removing the 'time' based multiplexing with a code-based multiplexing. A lower frequency band (800MHz) also assisted in range by reduced path loss and attenuation.
  • Picture a room full of people having conversations – under TDMA each person takes their turn talking (ie time division), conversely CDMA allows many people to talk at the same time but is the equivalent of each person speaking a different language, ie in a unique code. This of course isn't exactly how it works, if you want to know more there are some resources at the bottom of the page.

EDGE

  • Enhanced Data Rates for GSM Evolution. GSM introduced a GPRS based packet data network in 2001, with a max speed of around 60-80kbps (downlink), equating to a download speed of 10kB/s – slightly faster than dial-up.
    EDGE was later introduced as a bolt-on protocol (no new technology was required) increasing the data rate of the 2G network to around 237kbps (29kB/s).

3G – The Mobile Broadband Revolution

  • Introducing the 2100MHz network. Three Mobile in conjunction with Telstra brought the 3G standard to life in 2005, servicing major metropolitan areas initially and over the following years expanding coverage to 50% of the Australian population. Leased out to Optus/Vodafone/Virgin, the 2100MHz combined with a 900MHz network forms the basis of all non-Telstra mobile broadband services, servicing around 94% of Australian residences.
  • The 3G standard utilises a new technology called UMTS as its core network architecture – Universal Mobile Telecommunications System. This network combines aspects of the 2G network with some new technology and protocols to deliver a significantly faster data rate.
  • The base technology of UMTS is the WCDMA air interface which is technologically similar to CDMA introduced earlier, where multiple users can transmit on the same frequency by use of a code based multiplexing. Wideband CDMA (WCDMA) takes this concept and stretches the frequency band to 5MHz. The system also involves significant algorithmic and mathematical improvements in signal transmission, allowing more efficient transmissions at a lower wattage (250mW compared to 2W for 2G networks).
  • The new network also employs a much more secure encryption algorithm when transmitting over the air. 3G uses a 128-bit A5/3 stream cipher which, unlike A5/1 used in GSM (which can be cracked in near real-time using a ciphertext-only attack), has no known practical weaknesses.

So how is 3G faster than EDGE?

  • UMTS employs a protocol called HSPA – High Speed Packet Access, which is a combination of HSDPA (downlink) and HSUPA (uplink) protocols. The Telstra HSDPA network supports category 10 devices (speeds up to 14.4Mbps down) however most devices are only capable of category 7/8 transmission (7.2Mbps down), and its HSUPA network supports category 6 (5.76Mbps up). These protocols have an improved transport layer by a complex arrangement of physical layer channels (HS-SCCH, HS-DPCCH and HS-PDSCH). The technological implementation of HSPA will not be discussed here but for a basic explanation feel free to watch the below video.
  • The only major limitation of the 3G network is, not surprisingly, coverage. As stated earlier the 2100MHz network is available to around 50% of Australia's population and when combined with a 900MHz UMTS network available to about 94%. As expected, the higher 2100MHz component suffers far more attenuation and FSPL and is often considered a 'short range' mobile network which is why a lower 900MHz network is required to service many regional and rural areas.

Next-G – 3G on Steroids

  • To overcome the coverage limitations of regular 3G, Telstra introduced its Next-G network (considered a '3.5G' network) in late 2006, operating on the 850MHz spectrum. The lower radio frequency coupled with a far greater number of phone towers is responsible for Telstra's Next-G network being over twice the geographical size (around 2.2 million square km) of any other network, and servicing 99% of Australian residences.
  • Aside from coverage, the other major selling point behind the Next-G network is its blisteringly fast network speed. Rated up to 42Mbps (up to 5.25MB/s) the network has the ability to operate faster than the theoretical maximum of most high speed cable internet services. This is the result of an enhanced packet data network – HSPA+ which was implemented in 2008 as an upgrade to large portions of the Telstra network.
  • HSPA+ also known as Evolved HSPA, utilises Dual Carrier technology and 64QAM modulation order to deliver these high speeds. HSPA+ is responsible for the 'Elite' and 'Ultimate' series modems released in 2010, with the Elite capable of up to 21Mbps, and the Ultimate up to 42Mbps.
  • The Ultimate series modems theoretically double the speed of the Elite device by the utilisation of Dual Carrier HSPA+. This big increase in speed is achieved by the use of dual antennas, you can think of an Ultimate modem as having two Elite modems in the one unit. Combining this technology with MIMO "Multiple In Multiple Out" architecture we can hope to see speeds increased to 84Mbps (ie doubling the 42Mbps) on the Telstra Next-G network in the near future.

4G – LTE-Advanced

Initially available in major cities, airports and selected regional areas in October 2011, Telstra's 4G network offers significantly faster speeds, lower latency, and reduced network congestion.

  • The 4G network is based on LTE-Advanced – 3GPP Long Term Evolution. LTE is a series of upgrades to existing UMTS technology and will be rolled out on Telstra's existing 1800MHz frequency band. This new network boosts peak downloads speeds up to 100Mbps and 50Mbps upload, latency reduced from around 300ms to less than 100ms, and significantly lower congestion. For more technical details on peak 4G speeds check out our fastest 4G speed guide.

Most areas in Canada 4G has a 15MHz bandwidth and operates on the following frequency ranges:

  1. Tower Tx: 1805-1820MHz
  2. Tower Rx: 1710-1725MHz
  • 4G bandwidth (ie the width of frequencies we can send and receive on) is critical in supporting high speed and a high number of users. Because in order for your connection not to get confused with someone else's, each user is allocated a small sliver of frequencies that they can transmit on and nobody else can. You'll notice this most during peak usage hours, where as more people start using the tower it will reduce the width of your (and everyone else's) sliver of frequencies, resulting in each person getting a reduced download/upload speed.
    Naturally this is a very simplified explanation (for more info read up on OFDMA and SCFDMA) but for our purposes it will suffice.

5G – Newcomers

5G is a proposed, but the not-yet-implemented wireless technology that's intended to improve on 4G.

  • Some of the plans for 5G include device-to-device communication, better battery consumption, and improved overall wireless coverage.
  • The max speed of 5G is aimed at being as fast as 35.46 Gbps, which is over 35 times faster than 4G. However, data rates of tens of Mbps might be expected for thousands of users, and around 100 Mbps for metropolitan areas.

How long does it normally take to transfer a land-line number from Comcast to AT&T? vice versa from AT&T to Comcast?

I assume you are referring to the wireline providers. Going from mobile to mobile is much faster than going wireline/wireline or mobile/wireline.  When wireline is involved it may involve actual wiring changes. 

It depends on a few factors.  But if done smoothly should be no more than 3 days, and often shorter.  If you're in a rural area or are dealing with business lines it could add more time.

What is a cheap, but good, alternative to Polycom conference phones?

Have you tried out Sqwiggle? It uses the hardware built into your computer to provide really great presence with your team throughout the day. You can also start a video discussion with a single click. I'd recommend taking a look!

Sqwiggle – Remote Working, Collaboration and Communication

Full disclosure: I'm a cofounder

Are international long-distance rates falling as a result of IP-telephony-based competition?

Rates have fallen dramatically over the last 5-6 years. They are likely to continue to fall, but not as much, percentage wise. IP-telephony-based competition is only one driver. New entrants in the traditional space also put pressure on long-distance rates, as well as wireless and cable (which is at least partically IP-based, but not always thought of as such, compared to say Vonage).

What kinds of opportunities exist for someone with deep understanding in networks, graph theory, optimization, queuing theory, etc. who has applied this knowledge in telecom R&D for many years and is looking for new opportunities in which to apply these concepts?

Electronic design automation (EDA), especially EDA topics in digital VLSI design at abstraction levels starting from the logic/gate level.

Smart grids. As aforementioned.

Computer networking.

Networked embedded systems, especially with respect to different types of cyber-physical systems (see "Internet of Things").

Embedded/Real-time systems (and operating systems). Scheduling algorithms do make use of graph theory and optimization techniques. Queueing theory can model data transfer from memory to the processor, and between embedded/computer systems.

Computational biology, systems biology, synthetic biology, and bio design automation. Biological networks can be modeled with graphs. Optimization techniques can be used in the synthesis of proteins and cells. Communication between proteins/cells/neurons can be modeled with queueing theory.

Will SMS (i.e. text messaging) still be popular in 10 years?

Yes :).

Let me expand on that.

There are lots of exciting growth areas in mobile technology, from health and fitness apps and spoken search, through to Amazon Echo dot (house/personal assistants) and chatbots.

All of the above are connected to some extent with mobile technology, and will fuel the opportunity in this niche for years to come.

Whilst generic topics like SMS may be on a declining search interest trend, part of this is increased knowledge for searchers in the areas including voice search and chatbots (source – Google Trends):

When you look at the growth of SMS and mobile historically (source – Mobile and SMS through the Ages Infographic), it is easy to see how this can continue with associated technology.

How hard is it to replace a cellular network like AT&T or Verizon?

Firstly, I think it is misleading to look at the context of sunk costs and operating costs from the perspective of retail pricing. I think the points made in the question largely underplay the costs involved in establishing, and operating, and continually upgrading a wireless network.

As pointed out by M Mullineaux, running wireless operations is a *really* expensive affair. And to these operating costs, add the cost that carriers have paid to acquire spectrum and you have definite reasons why you can't wish away the carriers.

I understand there is lot of antagonism towards US carriers with respect to service and pricing, but for someone to think that a new player can come out overnight with an alternate network to solve all such issues is, IMHO, wishful thinking.

How big of an area and how many people does one cell tower usually cover?

It depends.

For WCDMA one sector-carrier (ie 5MHz for one sector) could perhaps handle 80-100 voice calls at once.

So a microcell (one radio, omni) might do 80-100 users in one circle;  a typical small macrocell (three sector, 1 carrier)  basestation would be 240-300 with three 'fans'; and a big one (six sector, 2 carrier) would be 960-1200 voice calls.

Data chews up capacity *much* faster. Theoretically, one user could be doing peak data of 21Mbps – and that would use a whole sector (so the big basestation could only have 12 of those data hogs!

In reality, the basestation (the scheduler) would 'ration' capacity. But even so, data uses a lot (a lot a lot) more capacity than voice, and correspondingly fewer users.

Those are the number of active, 'live' users at any one moment.
Many, many users will share that.
For voice that might be thousands of people covered (each one only makes a call occaisionally, for a few minutes).
Similarly for data: many users can share a channel (stat-mux) thinking it is all theirs, and, again, people only use it intermittently

The maths isn't easy but roughly each user is active a few percent of the time, so potential users covered = 10x-50X the maximum number of active users (thumb in air guesstimate)
(Lets skip over signalling and apps traffic for now;)

This is what is driving the need for small cells and femtocells: to serve all the data demand, have lots of smaller basestastions, each covering less area but giving more data to people in that patch.

Those are very rough estimates. There are so many real world effects – but it gives some idea.

Skype or VoIP is a lot less efficient.
See What is the data usage rate of Skype for iOS?
It can be 128Kbps.
For contrast, AMR (standard voice) is 12Kbps

So if everyone used Skype instead of circuit-switched cellular voice that small BTS would drop from 80-100 people, to less than a dozen at once (and the same 10% would apply to the number covered who coiuld use it, if they were to call).

If one or two of them were to do websurfing as well then the cell would be full.

This is one reason cdma2000 (Verizon etc) keeps voice users on one carrier, and data users on a different on ('EV-DO' – DO = Data Only)

Again, those are very rough estimates. There are so many real world effects – but it gives some idea.

How many Indian people use the mobile web?

There are multiple surveys on this issue that try to estimate a number. But if you are looking for a definitive source, then the quarterly performance indicators report from the Indian telecoms regulator, TRAI (http://www.trai.gov.in), should serve as a good starting point. iirc, per the latest report, just under 150 Mn mobile users subscribed for data access services in India.

However, you might want to take a closer look at the definition which I believe is for any subscriber who has activated data services atleast once, rather than current active users.