Re: Once upon a time detector vans existed
Pedant Mode (Sorry)
The vans detected the leaked local oscillator (not the IF) from the first stage of the radio receivers that picked up the TV signal. Colour TVs had more receivers to pick up the colour signals, and so could be distinguished from black'n'white sets.
The local oscillators themselves can be quite powerful (as these things go), around about 1mW, so they're easily detected in the street having leaked back through cheap mixers and up the aerial cable. The same thing still applies today for Freeview digital sets.
Bait
I ran a B&W set for ages acting as detector van bait, and always ignored the nastygrams accusing me of probably having a colour set (which I didn't). Saw the van a couple of times. Being an RF engineer and having access to some reasonably powerful kit, I was tempted to give them a nasty blast of a high power signal, see how they like that up their spectrum analyser.
Signal != Person
One of the problems I think they'll have with this new technology is that they cannot identify the people using devices.
I'll explain with the following scenario. I have a TV license, I'm entitled to watch BBC anywhere in the UK, including when I use an public WiFi network such as BT WiFi. I go to a friend's house, who has a BT hub. I use the BT WiFi that their BT hub has switched on by default. That friend has not got a TV license, and I'm watching BBC at their place but not on their private WiFi that comes from the same BT Hub. However the BBC cannot tell the difference; they're not allowed to examine the network packet contents, encrypted or not.
Another problem - two adjoined houses have their living rooms next to each other. The WiFi routers are in the same corner of the rooms, separated by only a couple of feet and the partition wall. One of the houses has a TV license, the other one doesn't. I bet they can't DF the emissions to the accuracy required to tell which of those WiFi routers is in which house.
The Courts' Dismal Approach to Science and Technology
My fear is that the BBC will be too gung-ho with prosecutions, and the courts will take an unreasonably optimistic view of the reliability of the technology. The UK courts haven't exactly been that clever at sorting scientific fact from pseudo-fact, and there's too many holes in this technique for it to be relied upon as the sole evidence required to jail someone.
The Courts have been appallingly willing to accept scientific evidence with low probabilities of correctness as being evidential fact. If an 'expert' states in court that something is fact then the court accepts that, and no amount of dissenting scientific opinion will change their mind. As defence you're not even allowed to challenge the "expert" evidence in court or even discuss probabilities.
This caused a number of people to be jailed on DNA evidence alone, until someone irrefutably showed that the number of base pairs being accepted at the time as "good enough" wasn't. A man accused and jailed for rape commissioned his own more thorough DNA analysis using his own money. This showed that it was a close match but definitely not an exact match - definitely not him then. Quite a few cases got quietly squashed as a result.
Misguided and Easily Circumvented
Effectively they are doing a primitive traffic flow analysis attack on encrypted communications. Well, that's easy enough to defeat in software. As the article suggests changing the network MTU would be one thing. But it wouldn't be hard to develop an app, or even a website, generating network traffic that'll bugger up the analysis too.