Technological - Telecom for beginners 2007

Basics of Electronic Communication

The importance of waves

All of the data transmitted in telecommunications is transmitted as an electromagnetic wave.

These waves can either travel down a guided channel, i.e. a fixed line, such as a fibre-optic cable, or they can travel through the air, i.e. wirelessly, such as mobile phone signals.

But what is wave? In general terms there are four key details describing a radio wave:

Wavelength

The wavelength measures the length of each wave; the distance from the start to the end of a wave. Each wave has amplitude, i.e. an individual strength, which is the value that will be recorded for it. In a digital wave, there will be two distinct amplitudes, with one corresponding to 1, and one to 0 (i.e. its binary coding)

Longer wavelength signals bend more easily around obstacles, so they will travel further than shorter wavelengths. As such a light-wave, where the amplitude is around 1 billionth of 1meter will not bend easily around obstacles (hence the reason we have shadows), whereas TV signals which has an amplitude around 1meter are more malleable and can therefore bend around obstacles.

Frequency

Measures how many waves come each second. Frequency is inversely proportional to wavelength, according to the formula Frequency= Speed/Wavelength. Electromagnetic waves travel at the speed of light (300,000,000 meters per second), so frequency would be 0.3 × 10⁹ / wavelength.

High frequency waves have high data capacity (bandwidth) and so can carry lots of data. This is due to the fact there are many waves, i.e. data-points, in a short space of time and each wave can carry a coding point (bit).

Strength

Stronger signals travel further as the wave will take longer to peter out. The downside is that they may interfere with other signals being transmitted elsewhere in the same frequency.

Analogue/Digital

Analogue signals vary continuously, so there is a value at each point, and analogue waveforms look smooth. People see and hear analogue signals.

Digital signals have discrete values, typically one of two different values at each data-point.

This data can then be interpreted by recording and processing a string of 1s and 0s, called bits (binary digits).

Figure 130: Analogue wave Figure 131: A 32-bit digital wave

Wavelength

Amplitude

Analogue wave 1 wavelength

1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 0 0 0 0 0 1 1 0 0

Digital wave

Source: Deutsche Bank Source: Deutsche Bank

Packets and switching

Telecom traffic (voice and data) typically needs to be directed along a network, like traffic on a railway system. Routers and switches sit on junctions in the network, and direct traffic along the right route, according to its destination, and their knowledge of the network.

Networks are most commonly built in loops so there are multiple means of getting to the end point. This allows for capacity management and is fails safe, ensuring the sustainability of service of a network element fails.

In a circuit-switched network, such as the traditional PSTN, when two users wish to communicate, a circuit or route is identified (by routers), and then held open all the way between them (i.e. bandwidth is reserved). This ensures constant quality of service on the connection, but is very inefficient. When users are not sending each other data, bandwidth is still reserved for them, and so remains empty. Using a modern day analogy circuit-switching is equivalent to running a marathon route that has been roped off so that people not racing are excluded from the running route.

In a packet-switched network, such as the internet (running on IP), when users wish to communicate, their data is split into packets, labelled with their source and destination addresses. Routers then direct the packets along the network towards the destination, using dynamic databases of the most appropriate route to each address. All packets travel together, fitting into whatever space (capacity) is available, and where excess space is available routers will identify a potentially quick route, so bandwidth can be used fully. This is equivalent to a mass of pedestrians walking around and consulting signposts when they reach a junction, with space never reserved in advance, but allocated to people on the basis of their occupying it at the moment. The randomness of packet-switching is its key advantage.

Because circuit-switching involves massively cordoning off bandwidth and preventing its use, whilst packet-switching uses it as needed, packet-switching is vastly more efficient.

However, packet-switching means that time to delivery is unknown, as it depends on how many others are using each portion of the route. This is a serious problem for time-sensitive traffic such as a voice conversation especially when it is important that the order of packets is reconnected on the correct order. To solve this, protocols such as MPLS may be used, which label packets according to their temporal priority, and then allow bandwidth to be reserved for these to run along a predictable circuit-connection that is part of a network where other

Figure 132: Services available by technology

Voice VoIP SMS MMS E-mail Browsing Broadband IPTV

(VoD) Video-calls Games

PSTN • • •

GSM (2G) • • •

GPRS (2.5G) • • • • • •

3G • • • • • • • •

3.5G (HSDPA) • • • • • • • • •

DSL • • • • • • •

Cable • • • • • • • •

Wi-Fi (802.11g) • • • • • • • •

Wi-Max • • • • • • •

Satellite • • • • • •

Source: Deutsche Bank

Technology: Traditional voice

There are multiple ways of carrying voice traffic; traditional PSTN networks, IP networks (VoIP) or on mobile network are three examples that are in people’s consciousness. We discuss mobility later in this Primer and therefore in this section we focus on voice, where it is carried on a fixed-line pipe of some-kind. However, this definition is determined by type of infrastructure carriage, whereas VoIP is possible on any data network, including mobile, and therefore a more appropriate split of voice may be technological (switched versus IP (VoIP)).

Voice traffic was traditionally carried on PSTN and mobile phone networks but with the move to packet-switching IP is becoming increasingly important (and price deflationary). However in many cases these networks run parallel and internet access technology such as cable and DSL may have a PSTN voice channel in addition to an internet channels. VoIP requires a moderately fast internet connection, and is unsuitable for narrowband connections such as dial-up.

In Figure 133 we show the evolution of wireline networks and one interesting conclusion is that there is increasing simplicity within network developments.

Figure 133: Technological evolution of wireline networks

Yesterday Today Tomorrow

Fibre Optics

SDH/Sonet Transport Voice -Circuit

Switching Data -Fr. Relay, ATM Access - DLCs, POTs, Data -Fr. Relay, ATM Access - DLCs, POTs,

Source: Deutsche Bank

Switching (circuit-switching; IP; MPLS) - detail

Network switches connect to each other to transmit information. Simplistically to send data from one point (node) to another a switch opens the channel between the points and then sends the data. In complex networks, the route between two communicating nodes will typically involve a string of such channels.

There are two important types of switching: packet-switching, and circuit-switching. In packet-switched networks, data for transmission is split up into discrete packets, which then

Figure 134: A structural mess: A 2003 attempt to map the internet; routers and switches sit on all these junctions and direct traffic

Source: The Opte Project

Basic small networks

Most nodes on a network have a MAC (Media Access Control) address providing a unique identity. Switches are increasingly intelligent and learn the MAC addresses of the other nodes connected to them, and when a switch receives information for a known MAC this becomes the preferred route. If the MAC is not recognised, the packet is sent to all alternative neighbouring switches to all neighbours save the sender. Switches are appropriate to small networks, whereby each node connects to the switch.

In a network consisting of two sets of computers attached to two different connected switches, each switch knows the MAC of those connected to it: firstly the set of computers, and secondly the other switch. If a computer attached to the first switch sends a message to one connected to the second, its switch will broadcast the message to all it other neighbouring switches hoping that other switches recognises the MAC and then redirects the message to the correct recipient.

If switches are only connected to end systems and other switches, every packet for a non-neighbour would propagate throughout the network, overloading it with duplicate misdirected traffic, so fail if packets are intended for destinations other than their neighbours.

The importance of routers

A router is like a switch but learns that there are other routes beyond its immediate neighbours, and therefore are able to connect multiple networks together. It can then use these routes to instruct switches. In an IP network, routers inform each other (either automatically or on request) about the networks they are connected to. Routers that receive this information record it in a look-up table, so that they know which of their neighbours can be used to reach particular systems, and they thus build up a picture of the network.

Each node is assigned a unique IP address, which is attached to packets to or from it, in the IP header. When routers need to send a packet, they consult their look-up table for the MAC they have recommended for packets to that IP (a certain portion of the IP will identify the host network of the node, and the rest will identify it within that network, so routers need only retain routes for host addresses, not routes to every individual IP). Routers can be

attached to multiple switches, and will then tell switches about which MACs to send packets to, based on their database of IP addresses. Switches thus send packets to other nodes on routers’ instructions, without knowing where they will end up.

Figure 135: A wireless router

Source: Telindus

The most widely distributed version of IP is IPv4, but is being slowly replaced by IPv6. The main difference is that IPv4 addresses are 32-bit long, compared to 128 bits in IPv6. 32 bits allows for around 4 billion unique addresses (~4.3×10⁹), which is too few for one per person;

whilst IPv6 allows for around 3.4×10³⁸, or about 4.3 x 10²⁰ addresses per square inch of the Earth's surface; plenty for every device to have an IP address. Having a huge amount of spare numbers also means that the system of assigning them can be tidier, much the same as in a system where telephone numbers have many digits. For example, if an IPv4 host network has many members, it may need to have several host addresses, in order to generate enough unique addresses for its members, and so this will generate multiple entries in routers’ address records for that single host, but in IPv6 it is easy to provide a host address that allows for plenty of user addresses in the network domain.

There are typically many routes along which a router could send packets for a particular destination, just as there are different routes and modes of transport one can take on a journey. IP itself doesn’t specify which route to choose, rather it describes the sending process; it is thus a routed protocol. A routing algorithm is required to decide which routes to take, based on things like speed and reliability. This then determines which MACs a router chooses when sending packets. A dynamic routing algorithm will constantly update the look-up table as it receives data about the network in order to achieve efficient routing. Data about the network is typically received via TCP (Transfer Control Protocol), which transmits data about IP transfers, e.g. when a router can’t pass on an IP packet it sends a message back via TCP to tell the originating system that the packet has failed. TCP is essential to ensure reliability, as without it there would be no way of knowing whether packets have arrived.

Circuit switched versus packet switched

In circuit-switched networks, bandwidth is reserved to a particular channel of communication, and so packets do not displace each other, but in a packet-switched network, the capacity available to a packet depends on what is being used by other packets. Here the traffic’s inherent unpredictability means that the speed of a packet’s arrival will vary dependent on network traffic. For applications where latency (the time for a single packet to traverse the network) matters; this is unacceptable. MPLS is a routing protocol that allows for the

label enables different routes to be chosen depending on the packet, e.g. so that those routes with constant and sufficiently low latency may be chosen for time-sensitive packets, whilst other packets are sent where bandwidth is greater. The description of the route in the label saves routers from searching for the IP in their look-up tables, thus saving time and computing for intermediate routers.

Figure 136: Packet-switching Figure 137: Circuit-switching

A B C

A B C A

B C

A B C

A A

C C

B B

A A

C C

B B

Source: eArchiv, Deutsche Bank Source: eArchiv, Deutsche Bank

Public Switched Telephone Network (PSTN)

This was the foundation of telecoms, copper cables that carry voice calls as analogue electronic signals, using circuit-switching. In the basic version, two wires are twisted around each other, with one carrying the signal, and the other reducing interference; after the design of Alexander Graham Bell. In the modern network, copper is generally the “last mile” into homes, with the main network carried over fibre-optic lines and cable. Though mobile phones connect to the PSTN, the networks of base stations that connect them into it are generally thought of separately. The PSTN is formally the concatenation of telephone networks. (This commentary is restricted to vanilla PSTN, leaving aside enhancements.)

Figure 138: Twisted Pair Copper Cable

Source: "Evolution of the Technology", Australian Photonics CRC, 1999

A universal technology

PSTN is literally worldwide, with just about every home in Europe connected. It varies in quality a little between countries, depending on age and maintenance. PSTN connects everybody potentially to everybody else: most homes have landline telephony, which links in directly to the PSTN. Apart from its main role for analogue voice calls customers can use a modem to dial-up through the vanilla PSTN to the internet’s packet-switched network via an ISP, but as internet usage matures, dial-up’s low-bandwidth (up to 56kbps) is increasingly inadequate. Numerous additional technologies (e.g. ISDN and DSL) have been designed to exploit the massive fixed resource in the PSTN network, to better the low speed it offers in vanilla form.

…but a legacy technology

The PSTN is essentially a legacy technology, which has been upgraded with other software and technologies to add bandwidth (especially compression technologies). However is most European and other developed count roes there has been little growth in provision of the basic offering.

It represents a large and integral asset for both those who own it, and those to whom they lease it (as required by regulators). As with infrastructure generally, expense can become more of an issue in remote areas, but even this is not generally a significant factor. Though the invested capital base was significant (Deutsche Telekom has a domestic asset base in its domestic network of Euro 27.8bn at the end of 2005), this is a sunk cost, and marginal costs are fairly small (often negligible) for many types of calls.

Marginal costs depend mainly on interconnecting and termination fees, whereby the service provider does not actually own the entire network involved. Costs thus depend on what connection is being made, and operators can match costs to revenues by creating pricing structures that encourage intra-network traffic.

Figure 139: Basic representation of a switched network

Source: International Engineering Consortium

Voice/VoIP

Voice traffic has traditionally dominated telecommunications and in most market fixed-line virus remains the dominant call origination technology. Value-added options to vanilla voice include services such as caller-identification and voicemail. There are also services offered via premium-rate numbers, such as tech-support, adult services, directory enquiries, telephone-voting, and conference-call hosting.

Traditional voice traffic is carried over circuit-switched channels, ensuring a constant speed of communication. Voice was revolutionised by the mobile phone, which turned a service that

Voice over IP (VoIP) is voice traffic carried as packet-switched data via the internet, rather than PSTN or mobile networks, taking advantage of its much cheaper bandwidth. VoIP requires either a normal point of internet access (e.g. a PC) equipped with a microphone;

speakers; and software, or else a dedicated device, which may be designed to look like a traditional phone, and which plugs into an internet connection.

Figure 140: UK fixed and mobile voice traffic volumes (bn of minutes)

174 173

166 167 164

34 43 51 58 62

0 20 40 60 80 100 120 140 160 180 200

2000 2001 2002 2003 2004

Fixed voice minutes Mobile voice minutes

Source: Ofcom

Fixed-line users will typically pay a fee to be connected to the network, and then per-usage fees, although there may be some services (e.g. minutes of calls) included in the fee. Call prices vary depending on who is called, as the operator must pay termination and interconnection charges. Third parties may offer services (usually cheap international calls), typically pre-paid, that enable users to route calls via their networks whilst on another service provider’s line. Revenue for premium services is shared between the telephone operator and the content provider.

VoIP technology bypasses the PSTN by going through the internet, thus saving interconnection charges. Mobile and normal telephones cost more because they must pay for access to the PSTN, with mobiles more expensive than normal telephones because of the historic cost of mobile networks and licences. All three voice technologies are networked with each other, but it is easier, e.g. to organise VoIP-to-VoIP, than VoIP-to-mobile. Note that the PSTN is a circuit-switched network, whilst the internet is packet-switched and thus much more efficient.

Figure 141: VoIP System overview

Customer Premises

PSTN user

PSTN Gateway

VoIP user

Access Transit

Narrowband Broadband

VSP**

Internet Core Transit

Access Customer

Premises

Router VSP**

Broadband DSL or Cable

Modem

ATA*

* ATA = Analog Telephone Adaptor, connects an Analog Telephone to a VOIP network

** VSP = VOIP Service Providers:the next generation telco

Source: Ofcom

Technology: Mobility

The electro-magnetic spectrum and the allocation of frequencies are key to the mobile industry, which is in the Ultra High Frequency (UHF) range. This allocations of spectrum effectively creates a barrier to access and a capacity restraint, whereas in the fixed-line arena capacity barriers are negligible.

Figure 142: The frequency spectrum

<3 Hz

>100,000 km

3-30 Hz 100,000 km - 10,000 km

3-30 Hz 10,000 km -1,000 km

300-3,000 Hz Super Low Frequency (SLF)

Communications with submarines

Very Low Frequency (VLF) Submarine communication, avalanche

beacons, wireless heart rate monitors FM and television broadcasts

Super High Frequency (SHF) Microwave devices, mobile phones (W-CDMA), WLAN, most modern Radars

Night Vision

Extremely Low Frequency (ELF) Communications with

submarines

Ultra Low Frequency

(ULF) Low Frequency (LF)

Navigation, time signals, AM longwave broadcasting

High Frequency (HF) Shortwave broadcasts and

amateur radio

Ultra High Frequency (UHF) Television broadcasts, mobile phones, wireless LAN, ground-to-air and air-to-air communications

Extremely High Frequency (EHF) Radio astronomy, high-speed microwave radio

relay

2G mobile services: GSM 900MHz, 1800MHz,

1900MHz (USA) 3G mobile services: at

c.3.2GHz