At the end of June we reported on Vodafone’s announcement of the Femtocell Access Gateway, an off-the-shelf product designed to provide 3G coverage for up to four handsets simultaneously, in areas of low or no coverage. Having taken delivery of our first release ready units, we have compiled our first in-depth test of the capabilities of the product.
Overview of test equipment
For this experiment the following equipment was used to test the effectiveness of the Vodafone Access Gateway:
Broadband connection – The minimum system requirement for the Vodafone Access Gateway is a 1Mbps broadband line. The property used for this test utilised a 4Mbps broadband line and was tested during peak usage hours.
Router – The router used to connect the Vodafone Access Gateway to was the Netgear DG834G ADSL2+ Modem Firewall Router. It was selected based upon its wide use and popularity in residential configurations. The Netgear DG834G was running the latest UK approved firmware release (5.01.14).
Devices – The Vodafone Access Gateway requires a 3G enabled mobile device. As the Adam Phones target demographic user of a 3G enabled mobile phone is likely to be a BlackBerry user, this was shortlisted as the most suitable device. The BlackBerry also has the added advantage of being able to run advanced diagnostic tests of the signal level achievable, which is a more accurate mechanism than the graphic depiction of ‘signal bars’. The BlackBerry 9000 Bold used in this series of tests was running the latest Vodafone approved firmware revision (188.8.131.521). In addition to the BlackBerry Bold, a Nokia E51 was also used to test the capabilities of the Vodafone Access Gateway and offer a comparison to the BlackBerry in terms of signal strength, call clarity and network attachment.
Connection of equipment
Connecting the Vodafone Access Gateway to an existing network is simple. Using the supplied Ethernet cable, connect the Vodafone Access Gateway to an available Ethernet port on the back of the router:
The gateway will power up by itself on connection of an AC adapter, although it will take up to five minutes to establish and authenticate with the Vodafone mobile network before it is possible to establish a signal on any devices approved for use with that specific gateway.
The gateway has four lights on the side which show (in order) power, internet availability, handset connected and cellular service availability.
Once successfully connected to the service, the broadband ADSL router will simply list the Vodafone Access Gateway as a connected network device and will assign an IP address to it.
The Vodafone Access Gateway uses an IPSec tunnel to communicate with Vodafone’s core infrastructure; therefore the need for additional router configuration is highly unlikely. Vodafone’s official documentation references that the gateway is specifically designed for ADSL network connections and is not suitable for office LAN environments. A business grade device will be developed over the forthcoming months to be used for such environments.
Understanding the results
Before presenting the test data it is important to understand the mechanism we are using to measure the signal strength achievable from the Vodafone Access Gateway. As previously mentioned, rather than relying on a graphical representation of signal from the home screen of a mobile device, these results were measured in dBm. Therefore it is crucial that we understand what number dBm represents good or poor signal coverage.
Absolute power of a signal is measured in wattage. The decibel system can only describe relative power – a gain of 3dB means your signal is twice as strong as it was before, but the dB scale fails to define where you are starting from or what your ‘zero’ is. Therefore, we specify dBm, indicating that our scale is relative to 1 milliWatt of power. 0 dBm = 1 mW.
The results presented in this study show signal strength as a negative value (for example -84 dBm). The reason you see negative values is that you are representing small but positive numbers, on a logarithmic scale. In logarithms, the value indicated represents an exponent. For example, under a log 10 scale, a value of -2 represents 10 to the -2 power, which equals 0.01. Likewise, a negative dBm means that you are applying a negative exponent in your power calculations; 0 dBm equals 1 mW of power, so -10 dBm equates to 0.1 mW, -20 dBm equates to 0.01 mW, and so forth. Therefore we quickly realise that it is far easier to describe a weak signal as -100 dBm as opposed to 0.0000000001 mW.
The final point to note is the lower the value the greater (better) the signal strength; -1 dBm is significantly better than -100 dBm. It should also be noted that the maximum output from a 3G/UMTS enabled mobile device is -33 dBm. As an average we would define signal strength in the following categories:
-105 dBm to -100 dBm (Poor signal, may result in dropped calls)
-99 dBm to -90 dBm (Poor signal, may result in call break up)
-89 dBm to -80 dBm (Average signal, unlikely to experience issues)
-79 dBm to -65 dBm (Good)
Over -65 dBm (Excellent)
Test sites & results
The Vodafone Access Gateway was setup and configured in the field test property and three locations were selected in the surrounding areas outside of the property to test the radius of signal output. Returning to our earlier coverage map, the red circled area on the coverage image at the start of this review outlines the property shown in the image below. The three numbered sections show the three outdoor test areas used as maxiumum distances where signal could be achieved. The red crosshair in the bottom right hand corner of the property shows the location of the Vodafone Access Gateway (situated on the ground floor).
Before beginning the test we further corroborated Vodafone’s coverage map by confirming that no signal could be achieved in this location, and in practice the closest signal coverage discovered was GPRS at a distance of 0.5 miles east of the test site.
Moving west of the Vodafone Access Gateway it was possible to achieve signal coverage at a distance of 40 metres and maintain a voice call, despite the signal having dropped to -120 dBm. Beyond this distance the signal was lost.
Moving east of the Vodafone Access Gateway, results were less impressive. 40 metres in this direction caused a total loss of signal, and it was not possible to establish a connection beyond 15 metres of the Vodafone Access Gateway. Looking at the satellite overlay of the test site, you will note two significant ‘patches’ of trees between the Vodafone Access Gateway and test location ‘2’, and we have attributed this to the lower distance signal achieved in this location.
Call clarity is an imporant aspect of the Vodafone Access Gateway, and it is important that the user experience is as close as possible to standard cellular sites. Our tests provided excellent results in this respect as the clarity of calls was indistinguishable from using any normal cell tower. Obviously, as the distance between the Vodafone Access Gateway and the handset increases beyond the -100 dBm threshold one must expect degredation of the call and, eventually, call failure.
Connection time from point of call initiated on the device to call ring/connect is again, no different to standard user expectations.
Email delivery via BlackBerry (connected to BlackBerry Enterprise Server 5.0) did appear to be slightly delayed by up to 20 seconds compared to the normal operating standard; this may simply have been an anomaly or could be the Vodafone hand off to RIM’s secure infrastructure taking slightly longer when the data is routed via the Vodafone Access Gateway. However, a delay of a few seconds in mail retrevial versus no coverage is an amicable trade off.
Impact of calls on ADSL
One of the less user-facing factors is how the Vodafone Access Gateway impacts the bandwidth availabile to the ADSL line to which it is connected. We ran a batch of tests which showed the bandwidth utilisation passes almost unnoticed (though it should be noted the ‘Handset in call on VAG’ test was using two simultanous voice calls on the gateway, and not it’s maxiumum of four).
While the Vodafone Access Gateway performs well in low/no coverage areas, its ability to reattach to a mobile device does vary. During initial tests, if the device lost connectivity to the provided cellular coverage at a distance of 30 metres, simply returning the device to within a distance of 30 metres would not automatically reattach the device and the gateway, and in some instances the device would need to be taken back to within 5 metres of the gateway before coverage was re-established.
Results in this respect varied depending on whether calls were in progress on other handsets connected to the gateway at the same time, and variable environmental factors. However, as the handset plays a significant role in establishing a connection to a fixed cell system (be it a fixed cellular site or femtocell based system), one must assume the majority of failures to reattach coverage will be largely attributed to the device and not the coverage emitted by the Vodafone Access Gateway (which remains constant).
The key remaining factor to test is the VAG’s ability to hand over calls from itself to a standard cellular site. As our test site provided no coverage in the surrounding area it has not been possible to test this specific element at this time, although Vodafone informs us that this will work without issue.
Another key question is how the gateway handles QoS. We know the gateway will provide whatever bandwidth the connected devices require (up to the maxiumum available via the ADSL line to which it is connected), but how this impacts other computers conencted to the same router cannot be accurately tested without a significantly bandwidth heavy application running on any devices connected to the gateway. As mobile applications are designed to run on devices with bandwidth as minimal as GPRS (56Kbps) QoS is unliely to become an issue for consideration.
Like any cellular site, the Vodafone Access Gateway has a limited range of coverage. As our tests have shown, the gateway cannot provide an absolute guaranteed coverage radius in all directions due to environmental factors and obsticles which may degredate the signal path between base station and device(s). That said, these factors are present in industrial scale cell site deployments, therefore seeing the same traits on a smaller scale is acceptable. It should also be noted that simultanious connection of four handsets to provide 3G coverage over an area as great as 40 metres is an impressive achievement for a device the size of a broadband router, which requires almost no configuration.
While we have not been able to test the hand over of active calls between a gateway and permanent cell site, Vodafone have informed us that calls will be dropped when a user walks in from the macro Vodafone network coverage to a zone covered by a gateway, however, calls will not drop when the user walks out of a gateway coverage zone in to the standard / wider Vodafone coverage network. On a related point, gateway to gateway handover is also not currently not available – Vodafone have suggested this may be possible in the future.
The other primary limitation of the service is that it relys on the user having a 3G enabled mobile device, therefore GPRS/EDGE and standard GSM handsets cannot benefit from the Vodafone Access Gateway.
Finally, it shoud be noted that while the Vodafone Access Gateway can maintain connection to a maximum of four handsets at once, the number of handsets permitted access to a single gateway by its owner is set at 32, and a handset can be configured to use more than one gateway (despite the lack of hand over).
At launch the Vodafone Access Gateway is, as previously mentioned, activated online with a primary handset and up to four additional numbers. Should a user wish to ammend the list of numbers allowed access to the gateway, customers will have to approach their service provider for such additions/amendments. In time Vodafone plan to have an online portal for its gateway customers which will allow them to administer their paired handsets online, but this service is not yet available.