What we used in our tests
A laptop running Windows XP
A Vodafone prepaid USB data dongle (Huawei K3265), which has a maximum rated speed of 3.6Mbit/s.
For femtocell tests, we connected to our new Vodafone Access Gateway from Alcatel Lucent. We understand this is rated at up to 6Mbit/s downlink and 1Mbit/s uplink.
For WiFi tests, we used 802.11b to a Draytek 2600 router with dual antenna.
The broadband service is a standard ADSL service with a line rate reported on the router as 3.36Mbit/s downlink and 0.448Mbit/s uplink.
The free speed test service from www.speedtest.net was used to measure downlink and uplink speed and latency from a server in the UK.
The tests were run in a different room from the femtocell, so the signal had to travel through two brick walls.
The table below shows our measured results
|Downlink||2.36Mbps||2.38Mbps||2.18Mbps|| 1.5Mbps peak
|| not measured
The outdoor cellular data service was not able to run the test to completion. Instead, the download rate increased rapidly to about 1.5Mbps and then quickly ramped back to a trickle, so that the test failed. This behaviour was repeated several times. This is likely to be because the system is engineered to give a rapid/fast performance for short bursts of data requests (such as when websurfing etc.), but limit sustained peak throughput for individual users on the same cell.
Maximum speed of the equipment was limited by the wireline broadband speed of just over 3 Mbit/s line speed, rather than the WiFi, broadband dongle or femtocell which are all capable of higher throughput. With the average DSL line speed in the UK being around 4Mbit/s (according to a report from Ofcom), this situation is not untypical.
Data performance using the femto is very similar to WiFi. This contrasts dramatically with outdoor cellular data performance, which varies tremendously based on location and use by devices sharing the same cellsite.
When used for standard web surfing, pages were returned quickly and provided a good experience, similar to WiFi. This included watching YouTube videos and similar material.
Perhaps surprisingly, there wasn't much difference with Wired connection which provided the best results.
For those considering use of mobile data services by femto rather than WiFi, it's very promising. With Vodafone's current offer, I had to pay them for the use of data sent through my femtocell (in addition to paying for wired broadband) where WiFi and wireline data would not incur any charges.
NB: I had previously reported variances in the "ping" times which I wasn't able to repeat consistently during additional tests. I've therefore removed these results from the report pending further results.