Carrier Sense Multiple Access/Collision Detection (CSMA/CD)
IEEE 802
– WaveLAN
– 802.11 b
– 802.11 b Frame Struktur
– WLAN Netzwerk
– WiFi
HIPERLAN und 802.11 a
– 802.11 a Physical Layer und Frame Struktur
802.11 g

Development of Wireless Networks

In addition to the mobile phone network, another wireless technology developed from the late 1990s onwards. However, it was not driven by telephony but rather by the need to connect computers and their peripherals, initially wired and later wirelessly. The best-known result of this development is WLAN, the most widely used technology for wireless data transmission to this day. It emerged from the so-called Ethernet technology.

Ethernet

Networks

Since DARPA’s work on what later became the Internet, the industry has been concerned with networking computers and their peripheral devices such as printers and storage devices. DARPA was initially concerned with networking computers that were far apart from one another.

In the course of development, however, there was also a need to network several computers (later especially PCs) locally, i.e. within a building.

In general there are different Network sizes to connect devices:

• WAN: Wide Area Networks
  These are networks that extend across entire countries, such as national mobile phone networks
• MAN: Metropolitan Area Networks
  These networks run across entire cities and their surrounding areas. The early cellular networks in the US were typically MAN.
• LAN: Local Area Networks
  These are networks within a building or a campus.
• PAN: Personal Area Networks
  These are small networks around end devices. E.g. Bluetooth Pico networks

At the end of the 1970s, the focus was primarily on the LAN. How to create a simple network of small computers within a building or campus?

CARRIER SENSE MULTIPLE ACCESS/COLLISION DETECTION (CSMA/CD)

In the beginning, computers were connected to each other or computers to peripheral devices in a point to point manner. So there was a connection in form of a serial or parallel connetion. However, in a network there should be a “common wire” or medium for transmission that all devices can use and over which data can be transmitted using packet switching technology. But how is it possible to share a medium?

An American engineer named Robert Metcalfe developed a solution for this in the 1970s: Carrier Sense Multiple Access/Collision Detection.

To understand this principle, imagine a group of people all sitting around a table. Two people can talk to each other at any given time. If two people want to talk to each other, they have to wait politely until the previous person has finished their conversation. As soon as there is a sufficient pause, the pair can start the conversation. However, it might get problematic if another couple has also been waiting for a break and starts a conversation at the same time. Now the couple or couples will immediately realize that they have disturbed each other and will immediately fall silent. After a short break, one of the pairs will start again and hopefully not fall into the other pair again.

The situation is similar with CSMA/CD. A device “listens” to the media to see if it is free. If it is unoccupied for a certain predetermined time, it starts sending on its own. At the same time, it immediately checks whether there has been a disruption, i.e. an overlap with another participant. Is that the case. It immediately interrupts its transmission and waits for a random time before attempting a new transmission.

Metcalfe called his system Ethernet, after the (nonexistent) ether that transmits electromagnetic waves. He founded a company called 3COM and contacted the leading companies in the computer industry, Intel, Xerox and DEC, to standardize his Ethernet.

IEEE 802

Standardization was addressed by the Institute of Electrical and Electronics Engineers (IEEE, Triple). The global IEEE, founded in 1962, had established itself as the leading standards organization worldwide. In addition to standardization, the IEEE also organizes leading conferences in various technical areas (39 different societies). A particular focus of standardization is in the area of computer technology.

Standardization projects or organizations were numbered consecutively at IEEE. When IEEE was approached by DECT, Xerox and Intel regarding Ethernet, the standardization was given the number 802. The goal of 802 was to specify Local Area Networks (LAN). This was probably the next available number but many also see a connection in the fact that the project began in February 1980.

The Ethernet specification originated in a subgroup called 802.3. In addition to Ethernet, two other methods were specified here, but they could not become established in the coming years. Ethernet was finally standardized as 10BASE5 in June 1983. It included the CSMA/CD access method and a transfer rate of 10 Mbit/s, which was very high in 1983. At that time, transmission was still carried out using coaxial cable. The Ethernet cable with its special connector, as we still know it today, first became popular at the end of the 1990s. (802.3i).

Wireless LAN

WAVELAN

National Cash Register was an American company that had been building cash registers since 1884. In fact, the founder of NCR had invented the cash register. NCR was innovative and managed to get into data processing in the 1960s and even supplied computers in the 1980s.

At the end of the 1980s, a small team at NCR in the Netherlands looked into the question of whether cash register systems could be wirelessly connected to a network. The team was inspired by the 2.4 GHz ISM band and new rules that the FCC had set for Direct Spreading Spread Spectrum (DSSS).

We have already discussed DSSS under CDMA. The aim of DSSS is to spread a rather narrow-band digital signal and thus make it robust against interference and interference.

At that time, Ethernet still worked at 1 Mbit/s. The engineers also wanted to maintain this bit rate for their wireless transmission. However, this meant that the basic data rate was already quite high and a rather short spreading sequence was chosen, namely so-called Barker codes. Barker codes, like Welsh codes, have very good properties when it comes to autocorrelation.

The NCR team took a Barker code of length 11. Each bit is spread with the code or its complementary code. This makes the transmission very robust against incorrect transmissions as many possible errors can be corrected. If you transfer the chips with a chip rate of 11 MHz, the data rate is exactly 1 Mbit/s.

Modulation was initially carried out using simple Differential Binary Shift Phase Shift Keying (DBSPK). This creates a signal bandwidth of 20 MHz. To achieve 2 Mbit/s, Differential Quaternary Phase Shift Keying (DQPSK) was used, allowing 2 bits to be sent per transmission.

NCR named the new wireless transmission system WaveLAN and brought it onto the market in 1990. A year later, WaveLAN was presented to the IEEE 802 group and a standard began to be defined under the name 802.11. This essentially took over the transmission of WaveLAN. NCR was acquired by AT&T in 1991 and sold WaveLAN technology under the name Lucent. (Lucent Technology was a later spin-off from AT&T).

In 1997, the first WLAN standard based on WaveLAN was published by 802.11.

802.11 b

After the first 802.11 release, work was done to expand the standard, especially with regard to higher data rates. The company Lucent Technologies in particular introduced Complementary Code Keying (CCK). CCK replaces spreading with Barker codes. CCK takes on the double task of spreading as well as additional encoding of the data bits. Two CCK modulations increase the data rate to 5.5 or 11 kbit/s.

Data rate (Mbit/s)	Code (Length)	Modulation	Symbol Rate	Bits/Symbol
1	Barker (11)	DBPSK	1	1
2	Barker (11)	DQPSK	1	2
5,5	CCK (8)	DQPSK	1,375	4
11	CCK (8)	DQPSK	1,375	8

802.11 B FRAME STRUKTUR

802.11 b sends all data in frames. These are sent out unsynchronized. There are three types of frames for management, control and data transmission. The structure of a frame is shown in the following figure.

A frame is transmitted in two data rates. Since the data rate for the transmission of the user data is not initially fixed, the first part of the Physical Layer Conversion Protocol (PLCP) frame is always transmitted at 1 Mbit/s. This consists of the preamble and a header. The preamble consists of a synchronization sequence, which is again generated by a PN sequence. Any “listening” device can synchronize with this sequence and synchronize the start of the frame. With a 16 bit sequence SFD (Start Frame Delimiter) the beginning of the following header is ultimately displayed. The following header describes, among other things, the data rate and the length of the following user data. At the end of the header there is a CRC to check correct data transmission. At the end, the actual data is transmitted in the PSDU (PLCP Service Data Unit).

WLAN Network

802.11 b allows different types of connections.

• Ad hoc:
  This allows two or more devices to establish a network/connection.
• Basic Service Set (BBS)
  A network is formed by several terminal devices and a central access point (AP) (access device).
• Extended Service Set (EES)
  In the EES there are several access points that are connected to each other and, for example, cover an entire building or campus. End devices can switch between the APs.
• Wireless bridging
  Here, several devices form a bridge to forward data.

The most common use case is the BBS. An AP forms a connection to the end devices as well as a connection to the (Ethernet-connected) computer network or the Internet. The AP is the central element of the BBS. All data transfers, including those between the end devices, run via the AP.

The AP selects a frequency when commissioning. In the 2.4 GHz ISM band, up to thirteen channels are available (depending on the country) with a bandwidth of 5 MHz. However, since WLAN requires a bandwidth of over 20 MHz, the bands overlap. For example, in Europe only 3 bands can be made available without overlap (typically bands 1, 6 and 11).

Once the frequency is selected, the AP sends so-called beacon frames. Beacon frames are sent every 100 ms and provide information about the AP. Above all, the name or identity number, the SSID (Service Set Identification) and properties/functions. When a device is switched on, it searches for beacons on all channels and lists those to the user. The user can select an AP and request a connection. If a connection is to be established, the device sends an authentication request. The AP responds positively and the terminal makes a connection request, which is also confirmed by the AP. From this point on, data can be exchanged between the end device and AP. Of course, at the beginning of this transfer there is the possible encryption of data or a “real authentication” by entering a key.

Each transmission of a frame is confirmed by the receiver with an “ACK” frame. If a frame is transmitted incorrectly, the transmission is repeated. If necessary, with a lower transmission rate to reduce the likelihood of errors.

After a device has registered and sent the first data, it does not have to constantly listen for an incoming frame. This would put too much strain on power consumption. Instead, it decodes every third beacon on the base station. The AP informs every end device in the network in the beacon whether it has a transmission in the input buffer.

WiFi

Although 802.11 was specified by IEEE, it was initially more of a niche product. What was missing above all was an instance that tested 802.11 b devices so that different manufacturers could use the same standards. Therefore, an alliance of the leading manufacturers of WLAN products was created. Including, of course, Lucent Technologies and Harris Semiconductors, 3Com, and Nokia. The alliance was initially called WECA (Wireless Ethernet Compatibility Alliance). In addition to compatibility testing, they also took care of a market name because 802.11 b DSSS was not a good name. So the term WiFi was invented, like Wireless Fidelity, based on HiFi High Fidelity. The name WiFi became so popular that the alliance was later renamed WiFi Alliance.

Apple was the first company to integrate WLAN into its products.

The first company to incorporate WiFi into a product was Apple. The iBook they offered in 1999 had an optional WLAN card. In addition to these cards, Apple offered a home access point it called AirPort.

Several semiconductor companies quickly developed WLAN chips for the new market. As a result, the WLAN solutions became smaller and smaller and could already be integrated into the first PDAs in 2002.

HIPERLAN and 802.11 a

It wasn’t just IEEE that had projects for the development of WLAN. The Europeans were also active in defining a WLAN standard. The standardization was organized by ETSI and the project was called BRAN (Broadband Radio Access Network). The system developed was called HIPERLAN (High Performance Radio Local Area Network). The first HiperLAN1 standard was defined in 1996. Like DAB, it was based on OFDM. ETSI had approved a band in the 5 GHz range (5120 – 5300) for this standard.

HIPERLAN used 52 subtends spaced at 312.5 kHz. This results in a bandwidth of around 20 MHz. So 9 parallel bands could be implemented in the 5 GHz band. A 64-point FFT was sufficient to synthesize the OFDM signal. By means of high quantization of the symbols in the frequency range (64 QAM), data rates of up to 54 Mbit/s were possible and therefore 5 times as high as for 802.11 b.

However, HIPERLAN did not become established as the standard. Standardization via IEEE had more support, especially from American computer and communications companies.

However, Lucent suggested using HIPERLAN’s OFDM-based access, the so-called physical layer, for an 802.11 standard. This idea was submitted shortly after the first 802.11 release, even before 802.11 b was defined. Therfore it was defined under the name 802.11 a.

Shortly before, the FCC, the American regulatory authority, had approved a 5 GHz band for WLAN applications, the so-called U-NII band (Unlicensed National Information Infrastructure). 802.11 a was specified for 5GHz and was not interoperable with 802.11 b. Operating at 5 GHz initially had two main advantages. Firstly, there were no competing systems in the same frequency band such as Bluetooth or the microwave oven and secondly, there were many more available bands and therefore higher capacities. However, these advantages came at the price of poorer propagation properties. As already discussed with cellular systems, the propagation becomes worse with higher frequency and there is more absorption, for example through walls.

802.11 a was passed in 1999 but did not catch on as quickly as 802.11 b. This was primarily because 802.11a was initially limited to the American market. In Europe they initially stuck to the HIPERLAN standard for the 5 GHz band. The two standards therefore blocked each other.

802.11 A PHYSICAL LAYER UND FRAME STRUKTUR

As already mentioned, 802.11a uses 52 subbands for transmission. They are spaced at 312.5 kHz and therefore occupy 16.6 MHz of bandwidth (so-called occupied bandwidth). A new OFDM band cannot be directly adjacent. There is a protection zone that widens the bandwidth to around 20 MHz. The subbands are numbered from -26 to 26. The “zero frequency” is not used. This is primarily for technical reasons, as an offset voltage occurs when the frequency is zero (see e.g. Direct Conversion Receiver). Four carriers (-21, -7, 7 and 21) are so-called pilot channels. They transmit with known symbols and can be used to correct errors in amplitude, phase and timing that can always arise between the transmitter and receiver from symbol to symbol.

With 802.11a, the data bits to be transmitted are corrected against errors using a Forward Error Correction (FEC) code. Depending on the channel quality, different redundancies are possible. 1/2, 2/3 and 3/4. (1/2 means 2 bits per data bit, 2/3: 3 bits for 2 data bits and 3/4: 4 bits for 3 data bits.

As mentioned, the I and Q channels in OFDM are amplitude modulated. There are 4 types of modulation. The simple Binary Phase Shift Keying (BPSK) with only two states, additional 4, 16 or 64 QAM, where I and Q are quantized in 2, 4 or 8 steps respectively. BPSK is very robust against noise, while 64 QAM is only possible with very good channel properties.

The pilot channels are always modulated with BPSK only. The possible data rates are:

Modulation	Coding	Bits pro Subcarrier	Bits per  OFDM-Symbol	Data bits pro OFDM-Symbol	Data rate (Mbit/s)
BPSK	1/2	1	48	24	6
BPSK	3/4	1	48	36	9
QPSK	1/2	2	96	48	12
QPSK	3/4	2	96	72	18
16-QAM	1/2	4	192	96	24
16-QAM	3/4	4	192	144	36
64-QAM	2/3	6	288	192	48
64-QAM	3/4	6	288	216	54

802.11 a therefore offers a range of data rates that can be adjusted depending on the channel quality.

Like 802.11 b, 802.11 a also transmits its data in frames (or bursts) of variable length. The bursts begin with two training phases so that the receiver can synchronize with the transmitter. In the first part, the amplitude, phase and time are roughly set. In the second part it is also possible to correct the channel distortions for the following OFDM symbols. Finally, an OFDM symbol ist transmitted to show which codes will be used and how long the burst is.

802.11 g

As I said, 802.11a and HIPERLAN blocked each other and there were only a few products for both systems. If at all, 802.11a was only used in professional areas e.g. in large offices or office buildings. Only 802.11 b quickly caught on because the 2.4 GHz band made it possible to spread worldwide.

The push for higher data rates for 2.4 GHz caused, that the physical layer of 802.11 a was also used for 2.4 GHz. However, it had to be ensured that this system was backwards compatible with 802.11 b. This lead to compromises, especially in the frame structure, which meant that the actually achievable data rate was lower than 802.11 a. The new high-speed WLAN was named 802.11 g.

802.11 g quickly established itself in the market, driven by the high data rates required by the new PCs and the Internet with DSL connections. DSL (Digital Subscriber Line) was used since 1999 and replaced the old Internet connections that were still routed via normal telephone lines. By the way, DSL is also a process based on OFDM.