Linux Troubleshooting for System Administrators and Power Users

3 Wireless networks generalities

Before describing existing WLAN standards, it is necessary to introduce some generalities applicable to the different systems.

3.1 Functions defined in WLAN standards

As for Local Area Networks (e.g. Ethernet), Wireless Local Area Networks standards specify layers 1 and 2 of the OSI model:

Layer 1, the physical layer, support the radio transmission service. It defines the transmitted signal (frequency band , channel bandwidth, modulation, filter, framing) as well as the necessary channel coding to ensure radio transmission robustness.

Layer 2, the data link layer, is sub-divided into two sub-layers:

• The MAC sub-layer support the media access service for the frame transmission. Depending on standards, this type of access can being supported with contention based or contention free schemes.

• The link control sub-layer is responsible for handling logical connections and interface with upper layers. Depending on the standards, the link control sub-layer may support the error detection and retransmission scheme using ARQ (Automatic Repeat Request) algorithm; admission control functions; connections setup and handling functions; radio resources control functions etc.

Hence, layer 2 support a transport service for data units delivered by the higher layer, i.e. layer 3 ("network layer" of the ISO model). Then, WLAN, technologies are commonly used to deliver IP datagrams over the radio link. However, in order to simplify implementation, current products offer the radio transport of Ethernet frames . This allows the delivery of a full service equivalent for higher layers, whether it is done over a classical wired LAN or a WLAN; the terminal protocol stack (e.g. the TCP/IP stack in a PC) will use the same internal interfaces (drivers) whatever the media.

3.2 WLAN architectures

Two types of architecture are supported for WLAN (depicted in Figure 12.1):

• In "centralised" (or infrastructure) architectures, wireless access is provided through an access point which manages the radio resources in a given cell . It permits access to the rest of the local networks through a "bridge" function implemented between the wireless and the wired LAN.

• In "ad-hoc" architectures, the WLAN is built over a set of wireless terminals in radio visibility range with each other which form a completely distributed system. This type of architecture permits the setting up of a network in a dynamic way depending on terminals which are in the vicinity of each other. It does not preclude the connectivity to a wired network as this service can be provided by a terminal supporting the two types of interfaces coupled with a bridge or a router function between the two networks.

Figure 12.1: WLAN architectures

Generally , WLAN standards are designed to operate alternatively in the two types of architectures.

3.3 Wireless terminals

It is foreseen a large number of wireless types of terminals: electronic pen, auricle, cellular phone, personal digital assistant, laptop, printer, web pad, digital camera and recorder etc. First are WLAN applications being developped in enterprises networks; WLAN products are currently oriented in this market segment and support PC interfaces essentially based on PCMCIA and PCI formats. With the coming of lower power consumption such as Bluetooth, some mass market product integration might appear.

3.4 Frequency bands

Globally, two frequency bands are identified for WLAN use: the 2,45GHz and 5GHz bands which, depending on continent and country, have different regulatory constraints.

3.4.1 The 2,45GHz band

It is the frequency band used for most current WLAN products. The total bandwidth is 80MHz (2400 to 2483,5 MHz). It is a "ISM" (Industrial, Scientific and Medical) band that can be used by any material conforming to electromagnetic compatibility standards. It is then not exclusively reserved for network operations, which implies that the system has to face important interference generated by objects of different types (such as microwave ovens for example). This band is available worldwide with some local restrictions in terms of emitted power or uses as summed up in the table below.

Concerning system channeling , 2 types of wireless techniques are foreseen:

• the DSSS (Direct Sequence Spread Spectrum) technology uses 14 channels of 22MHz with 5MHz spacing (ie there is some overlap between adjacent channels);

• the FHSS (Frequency Hopping Spread Spectrum) technology uses 79 channels of 1MHz each.

Bluetooth technology makes use also of a frequency-hopping transmission technique.

3.4.2 The 5GHz band

The following sub-bands are identified for being used by future WLAN systems operating at 5GHz: 5150 “5350MHz (worldwide use), 5470 “5725MHz (only open in Europe), 5.725 “5.825 (only open in North America). Globally, this permit use of up to 455MHz in Europe. However, the regulation allows only a sub-part of the band to be open , which should be of at least 330 MHz. Several WLAN systems (Hiperlan/2, 802.11a) are targeting the use of this band. However, they are based on very similar physical layers in order to permit economy of scale in chipset production.

The effective opening of these frequency bands is subject to local regulation in each country. At the European level, CEPT recommendations are foreseen:

• Full band allocation to Hiperlan/2 systems operating in indoor (max EIRP of 200mW).

• The possible use in outdoor exclusively in the higher sub-band (max EIRP of 1W).

• Sharing with radar and satellite systems using the 5GHz band is supported with the implementation of DFS (Dynamic Frequency Selection) and TPC (Transmit Power Control) which guarantee that the Hiperlan system will generate a limited interference level for the other systems.

Currently, in France, only the lower sub-band is open which allows only indoor use. The following table sums up the worldwide situation:

Compared with the 2.45GHz band, the 5GHz band is providing the following advantages:

• Higher bandwidth availability, permitting larger channelling (20MHz) and the coexistence of several networks with limited interference level.

• Spectrum sharing between a limited number of standardised systems and indoor use specifically dedicated to WLAN types of systems, which limit considerably inter-systems interferences .

This give the 5GHz band most attractive for applications needing high bit rates and guarantee Quality of Service. However, the current competition between the different standards as well as the European regulatory constraints may delay the worldwide market stabilisation for WLAN operating at 5GHz.

3.5 Range and capacity

Typical range for WLAN systems are of about 20 to 40m in a typical office environment and of up to 100 or 200m in Line of Sight environment. They are then relatively short, which is due to two main reasons:

• The emitted power is restricted both for practical reasons (battery consumption) and regulatory ones (power restrictions as seen in section 3.4); moreover, local area networks types of services are requiring high peak data rates, only possible with a good link budget, which limits the acceptable transmission attenuation and then the range.

• Those technologies were defined in priority for indoor private types of applications, ie not to cover outdoor extended areas.

In terms of capacity, current WLAN products support bit rates of 11 Mbit/s over the radio link, which permits really a useful bit rate of about 5Mbit/s at the IP layer. Emerging standards in the 5GHz band are targeting visent support of max bit rates of 54 Mbit/s at the physical layer. Lastly, WLAN systems are using time division schemes for sharing the radio resource, when a terminal is emitting it uses the complete channel bandwidth and then the associated peak rate.

3.6 Mobility

Terminals mobility between WLAN Access Points (in a centralised architecture) is managed by the terminals themselves that depending transmission conditions select the Access Point on which to associate. The handoff from one Access Point to another one is much closer to a cell "re-selection" scheme than a cellular handover controlled by the network as it is done in cellular networks. During this handoff , layer 2 connectivity is re-established. Then, when both Access Points are connected to the same local infrastructure (typically the same IP sub-net in a TCP/IP network), the network layer connectivity is maintained. However, when Access Points are parts of different sub-networks, the network connectivity can't be maintained as the terminal need to change its IP address and then start again its ongoing applications. In this case, in order to support a mobility service, it is necessary to use specific networks schemes, e.g. such as the implementation of the Mobile IP protocol. This mobility problem is not critic in enterprise networks which generally are based on switched Ethernet architecture and use routers only in splitting with the external Internet network. However, the problem may become more crucial for a campus size deployment where we may face to routed network architectures".

3.7 Security

Even if it is possible to implement security schemes in higher layers (e.g. by using the IPsec protocol at the network layer or end to end security at the application layer), the wireless link should not introduce security weaknesses in the communication system. Then, data prevention against eavesdropping as well as network protection against misuses access restraint the development of WLAN systems, particularly considering that radio propagation does not restrict waves to the user private place. Security functions defined for WLAN systems are:

• authentication preventing from network access of non authorised terminals;

• encryption preventing radio eavesdropping.

One of the main issue is to support a secured system for key generation and exchanges between terminals and the security manager (the Access Point or a centralised network server). Indeed, this is necessary as ciphering keys maybe broken when they are not revoked regularly. Two approaches are then possible:

• Secret key use which is based o the fact that each terminal owns a secret key, only known from him and the network and used as the basis for authentication and ciphering. The issue is then to provide a secure scheme to distribute secret keys in terminals and all the network elements implied in the authentication/ciphering mechanisms.

• Use of a combination of both public and secret keys: this type of approach permits a terminal and network elements to "publish" a key usable for another party to encrypt data that only the terminal having published the key can decrypt with an associated secret key. Considering that it requires more CPU, this type of mechanism is generally used only for authentication protocols and secret key exchanges to be used for data encryption. This type of scheme support regular revocation for temporary secret keys.