Home
Sitemap

TESTING EQUIPMENT FOR 3.5 GHz OFDM

A comparison between BreezeACCESS OFDM and Wi-LAN Libra 3000.

Background: a number of licenses for FWA (Fixed Wireless Access) on 3.5 GHz has been given to operators in Sweden. Networks using this frequency band will be built during 2003-2004. We have made tests on equipment fro two of the larger manufactures: Alvarion and Wi-LAN. Note that this is our impression after just a short period of testing. Some things we can say for certain, others we have to investigate further by putting the equipment into operation for a few months.

[the base station]

BreezeACCESS OFDM: 19" rack, 4 HU, housing power supply and up to eight radio cards. External powersupply 48 VDC is not included. Gives a very robust impression with good stability and eight fans that give quite some noise.

Wi-LAN Libra 3000: 19" rack, 2 HU. There is not much inside the box, and as a whole it gives a bit of a prototype feel. The DIN connector for the power supply does not have a locking mechanism. N connector for IF-cable assembly for 375 MHz. Outdoor unit inside a robust cast box. TNC connector for IF cable, N connector for antenna cable.

[client unit]

BreezeACCESS OFDM: Outdoor unit in a robust die cast box. LED for measurements of received signal strength works well. IF frequency 140 MHz permit very long cables to the indoor unit eg RG58 for lowest price. Robustly constructed indoor unit with 24 VDC powersupply, LED for error detection alarm, radio traffic, Ethernet traffic etc.

Wi-LAN Libra 3000: Outdoor unit inside a robust die cast box, with or without integrated antenna. Two versions with IF cable assembly and indoor unit or Ethernet interface, with special connector for do-it-your-self soldering, a bit fussy...

[performance]

BreezeACCESS OFDM: The channel width is 3.5 MHz (instead of 7 MHz on Wi-LAN), so there is room for eight channels in 28 MHz frequency bandwidth (instead of four channels on Wi-LAN). It is also possible to use a 1.75 MHz mode, but hardly of any interest.

The data rates are 2, 4, 8, 12 Mbit/s. At 4 Mbit/s we get efficiency with TPTEST (compared to 8-10 Mbit/s on Wi-LAN). The sensitivity in the receiver goes down -94 dBm at 2 Mbit/s and -79 dBm at 12 Mbit/s data rate. Fallback between the different data rates seems to work well. The sensitivity is thus up to 10 dB better than using Wi-LAN, which is useful, partly when using on really long distances or rather when having to take in to account altering propagation path with vegetation, seasonal variations and so on.

The output power is automatically set so that linkbudget is adjusted for every client. This we have not tested enough, but it seems to work. Configuration is done via a terminal, we have not checked if there are any graphical tools. As with Wi-LAN there are many setup options, and consequently many possibilities for making mistakes!

With two clients putting equal load on the access point, the bandwidth is distributed equally between the clients. We have not had the opportunity to make tests in a "real network, with real users". Wi-LAN uses a polled protocol, which is completely predictable. We think that BreezeCom uses CSMA/CA, so it is more the parameters like RTS/CTS and MIR/CIR that decide if the bandwidth is distributed between users in a correct way.

Overall the equipment gives an impression of being dependable.

Wi-LAN Libra 3000: Channel width 7 MHz, data rate up to 16 Mbit/s which gives 8-10 Mbit/s efficient bandwidth. It is possible to set three types of modulation 16-QAM, QPSK and BPSK, but there is no information on sensitivity for the different modulation types, instead it is stated as -82 dBm. The unit is administrated via a terminal interface. There fade margin is set relative to a -75 dBm input signal, which we interpret as a desired target value for linkbudget, and also corresponds to good performance in practical tests.

Administrating the unit is not entirely simple, many menus with options to set. For instance it is necessary to set the following parameters correctly, to be able to establish a connection; downlink frequency, uplink frequency, station id, sector id, sync id. Many menus give a "buggy" impression. At reboot there are long ping rates and many timeouts, before the link is up and running.

Distribution of bandwidth should be offered when the system uses a polled access method, but there are only functions for this in some clients.

[advantages BreezeACCESS OFDM]

- "feels better" than Wi-LAN, it can be so that Wi-LAN will shape up when the situation becomes serious. - can handle twice as many access points per site/cell for a given frequency band. - possibility for fallback, well even though there is a correct way of building networks, reality tends to alter things.

[drawbacks BreezeACCESS OFDM]

- less than half the efficient bandwidth, this I think will be the big bottleneck, especially as Wi-LAN has announced a 32 Mbit/s data rate for 10 MHz channel.

Links to the suppliers:
http://www.wilan.com (Swedish distribution Xaci)
http://www.alvarion.com (Swedish distribution UpGrade)

If you have questions, please contact Tonny Vilnersson or Niklas Gunhamn.

Smarteq Wireless AB, Box 4064, 182 04 Enebyberg, Sweden - Tel: +46 (0) 8 792 92 00 - Fax: +46 (0) 8 792 06 77 - Email: info@smarteq.se