* Posts by Andrew Somerville

16 posts • joined 9 May 2007

Facebook tosses creepy Place Tips beacons at stateside retailers

Andrew Somerville

They REPEAT don't collect any information

And you don't receive PlaceTips from the beacon.

A BLE beacon is a small box costing around £20. It is battery powered, typically with a life of 2 years between changes. As the name implies, it's a beacon - meaning all it does is periodically announce its existence using a unique ID. These announcements can be picked up any Bluetooth enabled device nearby. The beacon doesn't know anything about devices that have picked up their announcements and they are not Internet enabled so couldn't send any such information to Facebook anyway.

It is your phone that talks to Facebook's servers telling them it has detected a particular beacon ID. The server then checks this against a database of registered beacon IDs and hence knows where you are. Therefore for beacons to work with your phone they require (a) Bluetooth to be enabled and (b) the relevant app to be running (in foreground or background).

So if you don't want to be tracked, disable Bluetooth - but remember that every time you upgrade iOS, Apple sneakily turns it on again, or close down the Facebook app.

Wi-Fi was MEANT to be this way: Antennas and standards, 802.11 style

Andrew Somerville

Re: MU-MIMO - phones will still be as shit as they are now

Yep, generally they will be just as shit as they are now because they still have a single rubbish antenna surrounded by lots of other stuff.

The writer says "In our three-antenna scenario, it would allow the access point to use each antenna for a different device, or, perhaps one antenna for a phone and two antennas for a laptop." This is fundamentally incorrect when it comes to MU-MIMO. Think it through. In MU-MIMO the access point wishes to send different data to different devices at the same time. But in a typical access point all the antennas are omnidirectional and mounted in close proximity. So different signals sent from different antennas will normally interfere at the receiver. What is required to deliver MU-MIMO is beamforming - where the access point sends each stream from multiple antennas in such a way as to create a zone of constructive interference (for the wanted signal) and destructive interference (for the unwanted signal) at the receiving antenna. You need a minimum of two antennas to beamform. So an access point intended to serve two single antenna phones with MU-MIMO will need a minimum of four antennas not three.

In the real world we will have additional challenges to deliver the theoretical benefits of MU-MIMO:

* Any residual unwanted signal will reduce the signal to noise and hence lower the achievable data rate

* The access point will need to match up devices which have the necessary physical separation to obtain the benefits of beamforming

* The access point will need to match up devices that have simultaneous traffic flows

* MU-MIMO only works on the downlink (from access point to device)

* By splitting the transmit energy between multiple destinations you are effectively reducing the signal to noise ratio to individual devices. Again this will lower the achievable data rate

So MU-MIMO is probably only going to provide significant improvements when (a) we have access points supporting a large number of antennas and (b) the access point is supporting a large number of simultaneous devices.

Huawei: 'Tizen has no chance', Windows Phone is 'difficult'... it's Android all the way

Andrew Somerville

Re: 64,000

Only if you mispronounce it (like 99% of English speakers). The correct pronunciation is Wah-way

See http://www.youtube.com/watch?v=KGAqvpYLiGw

Anonymous wifi the latest casualty of Russia net neurosis

Andrew Somerville

Re: Not just Russia

In Italy similar rules apply. If you don't have a local mobile number you typically have to show ID to the hotspot operator to get a PIN code for access.

Ofcom saves piece of 4G spectrum pie for '4th operator'

Andrew Somerville

Re: More rip off again

Not an option. The reserve price on the 800MHz spectrum starts at £225 million

Service Birmingham offshores IT jobs

Andrew Somerville

IT staff on benefits

So all these back office staff in Birmingham are so hopelessly unemployable that they are going to be on benefits for the rest of their lives? If that's the case maybe its better that they aren't working on important IT systems.

AFAIK, ServiceBirmingham is a joint venture between the Council and Capita, so any cost savings are shared, hopefully resulting in lower council taxes.

Ofcom works out why Wi-Fi doesn't work

Andrew Somerville

Spectrum analyser - everyone should have one

You don't need to pay £199 for a Wi-Fi spectrum analyser. You can get the new AirView2 for £69 + VAT. See http://www.wifigear.co.uk/category.aspx?category=69C0CD86-2A9F-4A04-800B-F979965CC007

Westminster forced to switch off digital CCTV cameras

Andrew Somerville

C.L.U.E.L.E.S.S Philip Dagnan

Seconds of googling allowed you to find a camera that (1) was not available when Westminster specified their system (2) has a sensitivity of 2 lux so is a fat lot of use for seeing at night. (3) cannot stream full rate video.

Designing CCTV systems may perhaps be a little more complex than you think.

Andrew Somerville

Traffic Management Act 2004

Whilst the Act was published in 2004, it makes absolutely no mention of CCTV cameras. Instead it talks about "Approved Devices" that can be used for enforcement. The procedure for certification of approved devices was only published in early 2008, probably well after Westminster's system was specified. This 47 page document http://www.dft.gov.uk/pgr/roads/tpm/tmaportal/tmafeatures/tmapart6/certapproveddevices.pdf does mandate a minimum resolution of D1, but it also permits the use of digital zoom with a resolution up to 5 times worse than D1. So I fail to see why a solution which uses 4CIF resolution and optical zoom cannot be acceptable.

Cisco plays catch-up, gets directional

Andrew Somerville

Not a directional technology

As the Cisco description makes clear, this is not about them "getting directional" - it is about using the combined signals to/from multiple antennas to produce a more consistent signal that is less affected by multipath fading. The antennas used at the access point will have spatial separation, but in most implementations would be omnidirectional (just as in 802.11n MIMO deployments). In fact Cisco emphasise the benefits of this approach over a directional/beamforming solution.

Nevertheless this does not overcome three other fundamental issues associated with poor throughput in microcellular Wi-Fi, which in my experience generally have a greater impact on enterprise wireless performance - non-optimal AP selection by clients, co-channel interference and packet loss during roaming. All these issues are solved in Meru Networks' architecture.

Gates dethroned as tech's richest get thumped

Andrew Somerville

Missing comma(s)

No I think Ralph B is half right. What it should say to be grammatically correct, albeit with an unusual sentence construction, is "Gates, dethroned as tech's richest get, thumped"

Note: "get" noun, derog slang a stupid or contemptible person;

WiMAX has 'failed miserably'

Andrew Somerville

The network in Italy may work, but it ain't WiMAX

"Alvarion BreezeACCESS VL" is a product that uses a modified Wi-Fi MAC - there's not a trace of WiMAX in it.

There is no "5.4-5.7GHz" frequency profile in WiMAX

What do the "Motorola BTSs" do in this network? They certainly won't talk to Alvarion clients.

How do they get 10Gbit/s fibre to the top of mountains?

I'm guessing it's actually an Alvarion access network with Motorola backhaul connections at a few 10s of Mbit/s.

There is a world of difference technologically between a low density rural network serving clients with external directional antennas and a high density urban network serving clients with low gain antennas usually mounted indoors, although I would say that both have to be executed very effectively to make money.

Nokia N810 internet tablet

Andrew Somerville

Maybe people think it's a phone because ...

On the Nokia site http://www.nokia.co.uk/A4630128 it's listed under "phones" and the page has a section called "Get more from your phone with Nokia accessories".

US: Missile shield 'deactivated' until Iran tools up

Andrew Somerville

Shooting missiles at USA

They could shoot the missiles eitehr way round the world of course (if they actually had ICBM technology) - that's why the Yanks are also planning to have interceptors in Alaska and California where they don't need anyone else's permission. See http://news.bbc.co.uk/1/hi/world/europe/6720153.stm#map

Orange dismantles Bristol Tower of Doom

Andrew Somerville

UMTS macrocells

I'm afraid the poster has got his power sums completely wrong. I'm assuming he is getting his 27.9dBW figure from Ofcom's Sitefinder. This quotes power outputs in EIRPs (effective isotropic radiated powers), so is the figure AFTER all the antenna gains and cable losses. So the actual effective power in the main lobe is 617 Watts not 770 kilowatts.

In any case, his antenna gains are way out of whack. Firstly are the six antennas all part of the same array or are they covering separate sectors? Secondly are they transmit or receive antennas - receive antennas obviously don't count towards transmit gain? Thirdly, although 2 antennas will give an increase in gain of 3dB, you need to have 4 antennas to get 6dB and 6 would give 8dB - not 15dB.

Corporate wireless and alphabet soup

Andrew Somerville

RE: Can't overlap and others

I have to agree with others that much of Mr Kewney's article is a mix of the incomprehensible and the inaccurate (perhaps it was written by the other Guy?).

However on the 'can't overlap' issue he is generally correct.

Firstly with any vendor's system you can of course overlap the coverage of up to 3 11g access points because you have 3 channels that effectively do not interfere with one another. The problem comes when you have an office that needs more than 3 APs. You then have to reuse channels and in a 3 dimensional environment it is impossible to achieve the blanket coverage needed for telephony without coverage overlap (and hence interference) between APs on the same channel. Meru's single channel technology provides an effective solution to this problem.

Secondly, standard wireless access points are NOT viewed as a single access point. They may broadcast the same SSID, but they will be uniquely identfied by MAC address and clients will associate with a specific access point. This means that they need to roam when they move from access point to access point and unfortunately the roaming mechanisms in 802.11 are client initiated and pretty bad. By using a single channel and a common virtual MAC address, Meru shifts control of roaming to the infrastructure and hence delivers a much better user experience.

Biting the hand that feeds IT © 1998–2019