Clive Robinson • January 23, 2016 5:41 AM

@ Curious,

When I first heard about London having lots of surveillance cameras years ago I didn’t think much of it, but now, I can’t help but wonder if there are too many of them, as if this was some kind of mass surveillance.

I read the article a while ago and one obvious implication is that they were doing “mass surveillance”.

But the million dollar question is at what level…

The Met Police are known to be running their own equivalent of a Homeland Secret Service (think focus MI5) with agent provocateurs, spys and “black bag jobs” (they might even do “wet work” for all we know). They have certainly done an incredible amount of harm including illegitimately fathering children whilst doing under cover work.

The thing is that although they can get “phone records” they may not be of any use in identifing potential targets. It’s why the TAO catalogue advertised “Find Fix & Finish” tools (think about some of the El Chapo questions like “is who they have the real El Chapo or a double”).

Thus crossrefrencing phones via IMSI time logs and CCTV footage will potentialy reduce the number of candidates for identification.

Whist this may sound like the long and hard way to do this it has the advantage that involves few if any third parties outside of “the read in” group. Thus reduces information leakage to “tip off suspects”, provide trails for defendent defence investigation or worse still give journalists leads, which would reveal more of their immoral / illicit / illegal activities, primarily in support of “big business”, not to catch criminals or terrorists. That is ordinary law abiding people expressing their political rights through legitimate protest…

This of course leads on to the question of “coruption” in the Met Police at high levels… not the old fashioned cash in brown envelopes –though that does go on– but the early retirment into very cushy well paid minimal work posts. In some cases with the very people they were involved with the investigation of…

Clive Robinson • January 23, 2016 8:48 PM

@ Who?

May this be true?

Long answer short, Yes.

The way to attack symmetric crypto is by attacking not the encryption algorithm but the method/algorithm the keys are generated by.

The way to attack asymmetric crypto is via common primes or mass use of the same primes.

By common primes I mean those that have been selected badly often in embedded systems by poorly implemented Random Number Generators (RNG).

I’ve been pointing out on this blog for years now that we do not pay any where near enough attention to how we generate “random” and collect entropy. Thus if you were the NSA, GCHQ et al “Why attack the crypto algorithms when you can attack the RNG” or it’s lack of entropy.

There are lots of problems with not just the RNGs but how you use their output as well, and these are not well understood by system builders. Which is just one of the many sets of reasons why there are very many ways the likes of the NSA, GCHQ et al can find ways to by pass the crypto algorithms in working systems.

As I likewise often explain you need to seperate your encryption process from your transmission process. Using two seperate computers means two uncorrelated RNGs which breaks quite a large number of the available attack vectors… but as normal people rarely listen when they think that a single computer is both cheaper and easier to use…

As the old saying has it “You pays your money, and you takes your choice”.

Clive Robinson • January 24, 2016 4:20 PM

The BBC news has said the weather in DC is a not good with the metro closed and both airports badly disrupted and a meter of snow fall in a day,

http://www.bbc.co.uk/news/world-us-canada-35394742

With eleven states declaring states of emergency, it sounds like it’s not at all pleasent or safe. I hope any of this blogs readers in the effected areas are safe and well, and things get back to normal quickly.

Clive Robinson • January 26, 2016 12:18 PM

@ Bruce,

It would appear to be not just parts of the US going over board on “bomb threats” in schools.

The UK’s “Global Radio” 6PM news said sixty schools across England had been evacuated or closed due to “bomb threats”. Other news outlets give smaller figures,

http://www.independent.co.uk/news/uk/home-news/bomb-threats-spark-evacuations-at-14-schools-across-london-cornwall-and-west-midlands-a6834476.html

At the same time there have been almost as many severe storm warnings across the UK as the tail end of the weather that gave a meter or more of snow fall in the US over the weekend hits the UK. Some places that were recently flooded twice are being flooded again some by over a foot of water…

Clive Robinson • January 27, 2016 12:38 AM

@ JG4,

Couldn’t you use the radioactive circuit in a $10 smoke detector?

The problem with radioactive sources for such things as TRNGs, is firstly debiasing the half life. Secondly that the source is rarely pure, thus their may be several half lives to deal with which complicates the issue further.

Any physical source that varies with time in a semi-predictable way needs debiasing in time which is problematic at the best of times. The usual solution is to “not bother” and use a very short period of time to the half life. But you need to know this and be able to take accoubt of it if your experiment is to run for a long period.

@ Bob Paddock,

What about a Chaotic Oscillator

It depends on “the type of chaos” and the amount of “real entropy” -v- “faux entropy” there is.

It is fairly simple to design a RNG device that “close in” looks random. However when viewed further out it can be seen to have a regular pattern that is cyclic. Thus there is a lot of faux entropy and little or no real entropy.

An example would be two XTAL oscilators going into the CLK and D inputs of a D-type latch. Close in the Q output “looks very random” however further out you can see the transitions increasing and decreasing sinusoidaly at the difference frequency of the two XTAL oscs. The real entropy would be from the aging of the XTALs and by the time you debias it, it is measured in parts per billion in years, which is a very negligable amount of entropy.

Clive Robinson • January 28, 2016 8:35 AM

Kleptography the art of how to steal Keys

For some reason I don’t know this has poped up on Hacker news…

http://www.infosecurity-magazine.com/magazine-features/the-dark-side-of-cryptography-kleptography-in/

I’ve mentiond Kleptography a few times on this blog over the years (search for “moti” and “young”). But by and large most people don’t get it on their radar, even when it drops on their head through the cone of silence.

Personaly I think it’s something people realy should take more interest in as you can be sure it’s the sort of thing the NSA and GCHQ et al are upto their armpits in these days as stealing keys is way way easier than just about anything else for de-cloaking crypto protected data.

It’s also likely to form the basis of any “golden key” / “front door” backdoor that Comey and Co are prostituting themselves for.

Clive Robinson • January 31, 2016 11:08 AM

@ Thoth, Wael,

The question is “where to start?”, as I’ve said on a number of occasions you have to have a good knowledge of the whole computing stack, and thus a big chunk of mankinds understanding of the universe and himself.

So…

Security in all it’s forms is primarily about “information” which is a bit of a problem to start off with because it means many overlaping things to many people. As does unfortunatly “knowledge” and “random”.

If you take a very basic view of information it is fairly easy to see it is neither energy or matter, but can be easily impressed one way or another onto either of them. We are taught from a moderatly early age that matter and energy are effectively equivalent and constrained by forces and thus the speed of light. Einstein came up with his famous equation that couples energy, matter and the speed of light.

But information apparently does not get a look in untill you get to find out about what Einstein is reputed have called “spooky action at a distance”. Also “quantum teleportation” and “quantum entanglement”. Currently there is the question of the “What, how, why” of entanglement as previous assumptions come into question.

Increasingly it appears that unlike energy and matter information is neither local or constrained by forces… If true then amongst many other things ultimately their can be no security for information, only our inability to recognise it in a meaningful form due to impartial knowledge.

This should not come as a surprise as in essence it’s what the ideas behind the One Time Pad tells us.

There are three basic things we do with all information, communicate it, store it and process it. However the reality is we don’t do this to the information but the media on which it is impressed. That is we process not the information but the energy and matter it is impressed upon, and it is the constraints of the media not the information that limit how we communicate, store or process it.

Of the three processes communications is the fundemental one, as with a moments thought it can be seen that neither storage or processing can work without communication.

Thanks to the work of Claude Shannon we have a basic theory of communications in a physicaly constrained channel whereby energy directly or indirectly carries information from one locality to another. Shannon went on to show that with two or more channels it was possible to present an intermediary “eavesdropper” who only had sight of only one channel only partial information in that one channel thus secrecy could be achieved in that channel. However the valid sender and receiver having access to both channels had the full information and thus could recover the information.

Thus information security is about witholding some or all information from those not authorised to know it. If you are not authorised your task is to somehow gain sufficient information to make all information recoverable or atleast usefull in some way (ie via traffic analysis).

By and large humans are “simple souls” who perform tasks at some level sequentially. That is information is recieved as “input” processed in some manner and transmitted as “output”. This is the basic model used to explain all information handeling. Thus at every stage the model is a “data source” transmitting information down a Shannon channel to a “process” which transforms the information in some way and then transmits it down another Shannon channel to a “data sink”. Both the data source and sink are processes in their own right, thus we build chains of processes. The processes can be of two basic types “stateless” or “statefull”, to be statefull the process requires some storage to hold the state in. Which makes the storage both a data sink and data source and the process has effectivly two inputs, the current data and the current state held in the storage.

From this basic model any type of information processing system can be built and the examination of a CPU ALU shows this clearly. As does the rest of the CPU etc all the way up into the assembler language it interprets. Likewise from the assembler up through the high level languages and up the rest of the computing stack.

However the model is incompleate in that it assumes information goes only in one direction –left to right–, from input to output. It does not consider errors, exceptions and rate control.

Thus the model needs to be augmented to take this into account, and each Shannon channel actually becomes two the primary information channel (Left-Right) and a control channel (Right-Left) which provides feedback to a preceding proccess for rate, error or exception control / handling.

Thus a downstream (right side) process can send signals back to upstream processes (left side).

Which means there is a degree of reverse transparency back through the system.

Electronics engineers get taught about this when studying “small signal analysis” via “S-Parameters” and they get an apreciation of how such feedback can effect a system.

Importantly they learn that there is a time or phase component, that can effect system stability amongst other things as well as causing cross modulation and interference products. Computing systems wgilst not directly suffering from these problems, do suffer from their analogs.

Most “code cutters” never get to hear about such issues and rarely handle exceptions in sensible ways. Likwise errors tend to be handled badly or catastrophically in the “blue screen of death” or similar abnormal program termination. Worse exceptions or errors can due to bad handeling cause all the same sort of security vulnerabilities that poorly handled input can.

Even when code cutters do handle exceptions they usually.do it badly and errors tend to only checked for at the initial system input not where they will actually cause problems.

This pushing of error checking to the left has a number of side effects. Firstly it makes the input overly complex, difficult to debug and prone to “overly clever” optimizations that are very fragile and near impossible to upgrade or maintain effectively. Secondly, because the error checking is seperated from the code where it is needed for “code reuse” becomes problematical at best. Thirdly this alows the security to be bypassed from downstream.

To understand why this might happen consider the case of a partialy implemented control handler. The code cutter implements some kind of backwards data transmission to alow for say a “disk full” condition not to cause data loss. Thus the record that failed to get written gets passed back up the software process chain to a buffer pending a new disk etc being available. An attacker knowing this can fake a disk full but return not the actuall record but a malicious record instead. which would not have got through the error checking but has now bypassed it and is in the supposedly “checked data” stream to cause mischief.