c) And most of the testers are probably plastic dinosaurs.
44 posts • joined 15 Apr 2010
If you think 726 million scans a year is a lot, just wait until they've got the Bluetooth Covid tracing app out. That's going to generate around 2.5 trillion contact data points every year. Probably all stuffed into the cloud equivalent of a wet paper bag.
As others have pointed out, it's not going to work. Take the example of going to work in London after lockdown is eased. Bluetooth just loves propagating in metal boxes like the tube, escalators and buses, so you can expect 50 - 100 contacts to be logged on your trips in and out of work each day. If you've just caught Covid, you'll do that daily trip five times during the infectious, presymptomatic stage, which means 500 or more people will have you logged on their phone. On day six, when you self-isolate and get yourself tested, they’ll all get a message telling them to self-isolate and get a test.
With a working population of 6 million in London, we’ll see more than one new infection each day. If the infection rate is 0.01%, which is optimistically low, then that’s 600 infected people going to work, meaning 300,000 people with the tracing app will be told to self-isolate on day 6, or whenever the test results come through.
This is very Noddy maths. There will be quite a number of repeat contacts, as most people do the same journey each day, but that will be balanced by the new infections that join the spreading team on days 2,3,4,5, etc. However, it’s a good enough finger in the air check to alert you to the fact that we would need around a quarter of a million home-administered tests EVERY DAY. Otherwise, those 300,000 people sent home will be sitting around at home for three or four days to get their test results. Only a few hundred are likely to be infected, but if it takes four days for them to be cleared, the app will have locked down a quarter of London’s workforce.
It gets worse. Once they have the all clear and get back to work, it’s probably only going to be a few days before one of their new contacts is tested positive and they’re sent home again. After a couple of cycles of that, I can’t see users continuing to use the app.
Tracing and isolating is really important, but it starts with having massive home-testing available. A tracing app will make that much more effective, but you need the organ-grinder first, not the monkey, however much our politicians may identify with the latter.
It's the baked-in Government approach which killed the NHS patient data scheme and almost every other Government IT project / disaster, which is to impose central control. For more on that, read Richard Bacon and Christopher Hope's book "Conundrum - Why every Government gets things wrong."
What’s worse, is that it seems to be being designed by monkeys. When I spoke to some of the developers a few weeks ago, they weren’t aware of the fact that Bluetooth LE has been designed to be untraceable. So spurning the knowledge of a small group of experts within Apple and Google and the Bluetooth community who actually know how BLE works in phones does not appear to be a great design decision.
Here in London I'd take a guess that every NHS worker will come within range of someone with Covid-19 during the course of their working day, as will most bus drivers and other front-line staff. So they will all be warned to stay at home and self-isolate. Result - London shuts down and lots more people die.
The question is who does this actually help? The fortunate few like us, who have the privilege of being able to work from home, the front-line workers, or politicians desperate for a quick answer? Essentially, we're turning the knobs to balance Government spending, the economy and the number of people we bury. Anyone rushing into developing an app needs to understand that. This isn't just a case of tech going to save the world - it's tech being used to balance Government surveillance, long term costs and the death rate. I don't know where the balance is for that equation, but that's what we need to be debating, not which tech to roll out.
If you've not listened to Emily Maitlis' overview of who is really suffering from Covid-19, watch it now - https://www.youtube.com/watch?v=L6wIcpdJyCI. It's very relevant to this debate.
Strategy Analytics’ comment on the risk of Simcon’s reliance on mature technology seems to indicate that they believe the GSMA’s PR story about 5G a little too much. Looking at what’s happening in China, IoT growth is very clearly around the much simpler NB-IoT standard. According to various players in that market, around 10 million chips were deployed last year, with a prediction of over 180 million by the end of this year - most of them in modules. That means the overall volume of the IoT module market will double this year and most of it will come from Chinese vendors. After MWC I counted seventeen companies developing NB-IoT chips, which will probably feed into between 50 and 100 module manufacturers.
Compared with anything that has gone before, even 2G and GPRS, NB-IOT silicon is simple and cheap. That poses a problem for traditional, higher margin module vendors like Gemalto, Telit and Sierra. NB-IoT module pricing in China is predicted to be $3 at the end of this year and $2 by the start of 2020. It means that the bulk of the companies that Strategy Analytics are looking at won’t be able to complete and will probably look higher up the food chain to 5G, where they have more scope to add value. But that’s not going to have much to do with IoT. If they do retreat from NB-IoT, then the consequence is that China effectively owns the comms portion of the IoT. Which means that Western companies need to concentrate on developing positions higher up the IoT value chain. It’s not just about numbers - it’s about what you’re doing.
I suspect there's a clue in the fact that it's marketed as LTE, not 4G. LTE doesn't support native voice, but relies on using VoLTE. If you look at the supported networks list, they're the networks who currently have VoLTE support up and running. 4G normally implies fall back support for 2G and 3G; LTE doesn't. Hence my guess is that there is no legacy support for 2G or 3G and they're playing some clever tricks with voice packets over LTE. That should also help them with power consumption, as VoLTE's much more efficient.
As they say - they're sticking to their core strengths, of which comms is not one. So it's natural that given the need to choose between multiple unproven options, they go for the only one one which offers an SLA. What's surprising is to see a utility make a sensible decision when technology's involved.
It seemed a little ironic that the message on my Lumia saying that the app would not continue to be supported showed up under their strapline of "Maps for Life".
But I guess they're probably using the marketing experience from the VW group. Another one for trading standards and the ASA.
I remember a presentation from Sebastian Conrad - the lucky man who designed the well rounded Nigella range of kitchenware, where he recalled that when he was studying at the Central School of Art and Design in the late seventies, the Industrial Design course was known as the Department of Rounded Corners. Sounds suspiciously like prior art.
I think Andrew's answer says it all - they're small enough to sew onto clothing, bluetak onto lego, glue to stuff, etc, which makes them a lot more appealing as they then become a project about coding for something you already own. Much as I love the Pi, I suspect the vast bulk of users are of the age that grew up with BBC micros, or even Sinclair calculators. Or in my case log tables and slide rules. And the Pi is more for new projects, rather than adding functionality to something you already use.
I do think that adding Bluetooth to the device is a valid reason for not going with the Codebug, as it ticks the other box of letting kids connect it to their phones.
It's a shame about the power supply delay, but they're not the first to hit that problem. I've seen too many other product launches hit the same problem. The part everyone thinks is easy turns out not to be. Although numerous shocks with stuff I built as a kid never seemed to do me much harm. It did teach me that not insulating mains terminals is a bad idea.
The Smart Energy GB response is interesting, as it illustrates how their main interest is remaining funded as a quango, even if that means screwing the consumer. Sacha Deshmukh’s opening riposte is that “The Institute of Directors wants to reverse the modernisation of Britain’s energy system and take us back to an analogue dark age”. Had he spent time looking at the technology in these smart meters he’d have realised that they are the dark age. They were specified before the world saw the iPhone or had grasped the concept of the Internet of Things. The IoD’s main complaint is that the current smart metering programme is effectively a dark ages of smart metering technology. Out of date, over-priced and incapable of delivery any of the benefits which DECC claim. And the world is still waiting for DECC to release its financial justifications, resisting Freedom of Information requests to make then public.
There are important benefits that smart meters provide. The problem is that the UK deployment won’t provide them. The IoD report is a timely warning that DECC and the industry is sleep walking into another Government IT disaster.
There are some good reasons for a new protocol for home automation, as most of what we have is trying to adapt the needs of devices to it, rather than vice versa. Whether Thread goes far enough remains to be seen, but at least the folk at Nest come from an embedded device background which should give them a better perspective.
I do think it will kill ZigBee (http://www.nickhunn.com/is-google-and-nests-thread-a-zigbee-killer/), which will have some other ramifications, not least that we'll soon be deploying 50 million smart meters with an obsolete wireless standard that connects to nothing.
There was an interesting perspective from a senior manager at a US energy supplier earlier this year. When asked what the best thing about their smart meter roll-out was at an industry conference, he claimed it was the fact that the data gave them a greater ability to blame the customer.
The industry has been toying with Bluetooth for some time. Unfortunately they got side-tracked by ZigBee's PR and have spent a few years playing with RF4CE. That has given Bluetooth time to get its act together with Bluetooth Smart, which should start appearing in consumer products soon.
The good news for TV manufacturers is that Bluetooth Smart comes as standard in most Smartphones, and both Apple and Android have released APIs for developers. So by switching to Bluetooth, the industry sees the prospect of a future where they no longer need to include any remote control.
We start off with 33,000 malicious emails a day, but by the time Chloe Smith starts to wax lyrical halfway through that's down to 33,000 a monnth, or just 1,000 a day. Even assuming that users only get one malicious emial per day, which is very low, that implies that only 1,000 people have government email accounts.
So someone either needs to tell George Osborne about an unexpected level of cuts, or take Chloe out of the kindergarten and tell her something about the real world. 1,000 phishing emails a day is not DOOM. We don't need CUTTING EDGE TECHNOLOGY to counter it. But it would be nice to have a Government minister that doesn't descend into braindead mode the second they encounter a number bigger than ten.
Out of interest, I wonder how the cybersecurity folk like being lumped into the same industry sector as G4S security guards. Although now we've got 26,000 of them, there's probably enough to hold a red flag in front of every government PC.
I worked on rotating anode X-ray generators many years ago, which were very similar, in that they had a large mass of metal spinning at high speed in a vacuum. The biggest problem was maintaining the vacuum across a bearing that's rotating at high speed, whilst supporting a large mass. This won't need as high a vacuum as an X-ray generator, but even with the progress in ferrofluidic seals you'll still need pumps running to maintain the vacuum, contributing to the losses.
The other think he'll need to think about is the housing. When the bearing fails on a flywheel this size, the flywheel does a scary amount of damage, like ripping through several inches of steel. So you'll end up with something that probably weighs in excess of a ton. Which means you might be better off sticking with batteries.
At least DNLA understood that you need to specify conformance tests if it's going to work in the wild. Which is a rather important detail which is stangely missing from most IEEE specifications. As a result they tend to flounder in the market until an external body like the Wi-Fi Alliance comes along, takes them over and adds the missing bits.
So I'd conclude that it does look like DNLA, but without any interoperability or testing. Perhaps it would be wise not to ask Santa for any 1905.1 stuff for the next few decades.
There was a fascinating presentation at a recent Cambridge Wireless event by Laurent Simon of the Cambridge University Computer Lab, who not only pointed this out, but also the fact that it's pretty easy to tell whether the user is male or female, as you get very different accellerometer signals depending on whether you carry your phone in your pocket or your handbag. And it doesn't take much imagination to realise there a lot more you can pick up about what the user's doing.
You can download his presentation from http://bit.ly/WDpWgI
The marketing premise of these energy-havesting light switches is that because they need no batteries or wiring, you can just stick them where you want with velcro. Which means that you can have lots of fun by moving them all around someone's house, so that they appear to randomly turn on lights in other rooms.
Excellent for drunken parties. We need more standards like that.
Back in 2000, the Bluetooth SIG put a lot of effort into lobbying regulators to harmonise the 2.4GHz band around the world. That’s benefited the industry as a whole, as it means that wireless designers can design one global product. They never succeeded in a complete harmonisation, and the biggest gap is because of the Band 14 restrictions within the US, which means that Bluetooth, Wi-Fi, ANT and ZigBee products have a little less spectrum to occupy than they do elsewhere in the world.
Globalstar’s attempt to carve out alternative uses for the satellite band seem rather disingenuous. There’s a good argument for the FCC to ask whether the top end of the band actually needs to be protected, but consider following the rest of the world and make it unlicensed. In this light Globalstar are playing the pre-emptive marketing card of protesting that if their request is not granted they will take away internet access from our children, kill our kittens and put horse meat in our burgers. I’d side with the argument for increasing the unlicensed spectrum.
As a small start-up I quite like it. I agree that it will encourage trivial patents, but if I need to write a trivial patent to get some tax relief I'd claim that's where the innovation bit comes in. Whilst I'd pay a patent attorney for a patent that covers a real invention, I don't see the same need for a trivial one just for patent box. So it shouldn't cost more than £250 to get something granted that gets me tax relief on something I sell.
I just see it as a useful extension of R&D tax releif once I get the product into manufacture. But anyone who pays consultants or patent attorneys to get that trivial patent is probably wasting their money.
If the school isn't capable of running a timetabling program, it's probably not the best place to be taught A level ICT. Maybe your son should offer to do the timetabling for them in the future, as it sounds as if he could do a better job. And the drama course should help him deal with next year's awkward parents if their little darlings still don't get the options they want.
Has anyone worked out what the additional power consumption of all of thse multi-processor PCs running full tilt for 24 hours a day will be? Add those all up and from an energy perspective it probably make the Japanese Super K look like a solar powered BBC micro.
Still, if there's anyone out there, it should mean we're glowing nicely.
Although handset antenna design is very much black magic, there are some well known things to test when you’re designing a handset. One of which is that the antenna doesn’t work in the way the data sheet claims. As soon as you hold it, or put it next to your head the performance changes. That’s because parts of the body act as a ground plane, effectively detuning the antenna.
As a result an antenna that works well on a bench may barely work when next to your ear. That’s a phenomenon that has been exploited in the past to tune antennae so that they work better when you’re looking at it in your hand, rather than when it’s held next to the ear. The cynical reason for that is you’ll see a better signal strength when you’re looking at the display and consumer research has shown that if the signal looks good, users blame the network, not the phone.
That means a smartphone has a compromise: is it more important that you get good reception when showing an app to a friend, or when you’re trying to make a call.? If you always use a Bluetooth or wired headset then handheld is easy to optimise. Otherwise life gets very difficult.
Of course, the low power Wi-Fi approach is coming along just as Bluetooth low energy is launching, with its promise of reducing power consumption by several orders of magnitude below anything WI-Fi Direct can achieve. So this could be a classic example of building a brand new stable after the horse has bolted.
As always, wireless standards never meet their delivery dates. Wi-Fi direct was promised to be complete and have a qualification process in place around now. The Ozmo announcement implies that is probably still a year away. Although equally delayed, the Bluetooth low energy standard was published last December. Its qualification process should be in place by the end of this month and at least three companies are ready and waiting to start shipping chips.
One interesting point is that much of the Bluetooth low energy work is happening in Europe, which is frequently off the radar for US technology companies. As happened with GSM, they only believe what they see in sunny California, and then wonder what's hit them when the European standards arrive. So Bill McLean’s confidence may yet turn out to be more than a little myopic.
And to answer the pricing question, a Bluetooth low energy chip will be around one fifth of the cost of a Wi-Fi Direct chip, as it’s been designed for minimum silicon area. Plus it has the advantage that the volumes are higher, as they ride on the mass volume cellphone market, whereas Wi-Fi is limited to the much smaller smartphone and laptop markets.
There’s more below the covers of Bluetooth 4, or Bluetooth low energy as it’s colloquially known, than is implied in the article.
The range isn’t a result of higher power, as that wouldn’t work with ultra low power devices that run on coin cells. Instead there have been some fundamental changes to the radio specification, such as increasing the modulation index. It means that at a lower power of around 1mW you should get a range greater than 50 metres. There’s a good example of that being put to use in a radio controlled car at www.bit.ly.btlecar.
Bluetooth low energy’s focus is around consumer devices that connect to the phone. There is certainly a contingent of ageing geeks with an increasingly vested interest in health devices, but there are plenty of opportunities for a younger market. The spec includes the ability to measure distance between devices, so these range from child and pet monitors through to access control and lost phone detectors. And it will be cheap to put it into all sort of everyday products that connect via your phone to your Facebook page. So, for example, every time a Register scribe fills up their Bluetooth low energy coffee cup after a gruelling session at the keyboard, their public caffeine score can be updated for the world to see. You can see some more applications at www.bit.ly/btlespec.
As to who will win, that comes down to whoever gets incorporated into the most compelling products that people want to buy. Bluetooth low energy has the advantage that it will connect to a mobile phone that supports the new standard, and there are around 1 billion Bluetooth phones shipped every year. So whether it’s working with an app on the phone, or using the phone as a gateway to a website, the receiving half of the radio link exists. That means people can just but their new Bluetooth low energy toy as an accessory. That’s a major advantage that ZigBee doesn’t have – it has to persuade consumer to stump up for both ends of the connection.
The U.S. allows higher Wi-Fi powers than Europe, not lower. You can transmit up to 1W, compared to 100mW in most of Europe (France excepted). However, that's probaly a red herring.
All of Europe and many other countries require any new wireless transmitting device that works at greater than 10mW to be registered before it can be shipped. In Europe that's through the R&TTE approval process. It's generally a paperwork exercise, but it takes three months. That's something most manufacturers understand and incorproate into their product release plans. If Apple hasn't done it, it's either because it's forgotten, or not actually believed that anyone would consider the iPad worth carryign around with them.
Biting the hand that feeds IT © 1998–2021