Your browser is not supported. Please update it.

20 March 2014

Verimatrix declares victory for software security

While Elemental and Envivio have flown the kite for software based encoding, Verimatrix has done the same for content security. The outcome over encoding remains unclear in the light, for example, of Ericsson’s recently unveiled Virtual Encoding Platform, advocating dedicated hardware for high end broadcast applications. On the CA side however Verimatrix has now claimed an absolute victory for software in the multiscreen era, although with the ironical caveat that this victory has been achieved with the help of hardware, in particular SoC makers for set tops and client devices.

This has enabled Verimatrix to bury the old argument that software CA is inherently vulnerable to attacks that exploit the ability to access the code directly through debugging tools and analysis methods, and that therefore some kind of black box was required. Verimatrix can claim firstly that black boxes themselves were commonly attacked successfully and so needed constant refinement.

The main point is that the threat landscape has changed over the last five years as pay TV piracy has entered a new era, with defenses against attacks that extract plain control words from set top or gateway chipsets proven to have failed anyway, creating the need for stronger hardware security cores. The black box will no longer be a whole SoC but a secure region within that. This inner ASIC core will be protected not just from physical tampering, but from a growing variety of attacks that do not require direct access to the box but exploit variations in electromagnetic signals that can be used to deduce code words.

A common method of attack is differential power analysis, which enables key information to be recovered through statistical analysis of the chip’s power consumption profile and this can cope with obfuscation techniques such as insertion of random noise designed to combat the problem. Counter DPA measures therefore have to be more sophisticated, such as transformations to keys via some additional algorithms that are not affected by the cryptographic processes; and hardware techniques such as randomly varying the chip’s clock frequency to confound the DPA analysis.

Such techniques have been developed by San Francisco based Cryptography Research, which for that reason is a partner of Verimatrix in working with makers of set tops and device chipsets to embed the former’s CryptoFirewall ASIC security core technology into end to end security systems. To date CryptoFirewall has been deployed by some operators in conjunction with smart card based systems as an interim step, but then the actual pairing between the card and the SoC itself becomes a point of vulnerability subject to potential attack. Therefore both Verimatrix and Cryptography Research argue that it is better to dispense with card based systems altogether and that a pure security core approach actually gives stronger protection.

Whether you would call it a hardware or software approach is a moot point since like all good security systems it combines both. If the definition of a hardware approach is the use of silicon dedicated to a particular task, the distinction is still not clear, since a security core is clearly dedicated hardware while the rest of the system is software running on generic silicon. This suggests that particular debate may soon be put to bed as far as CA is concerned.

But there is another aspect here, which is that the major threat is most recently shifting away from control word access of any kind towards content redistribution of assets that may have been legitimately obtained. In that case no amount of CA security based on encryption will protect premium assets and the only recourse is to track down any sources of abuse. This is where digital watermarking comes in, with Verimatrix having developed its own technology called Videomark after being early to recognize its importance for content security. This is a forensic watermarking technology designed purely to combat privacy by embedding code that is invisible to the eye within the video stream at any chosen part of the distribution cycle that is under the operator’s or broadcaster’s control.

The idea then is that any camcorded version of say a premium movie in HD resolution would be traceable in principle to the screen where the copy was made, although we have yet to see any data pertaining to how reliable or consistent this process has proven in the field. The idea that video leaves a trace across its complete chain of custody is a good one and if seen to work will at least prove a deterrent.

As Verimatrix points out, the rise of content redistribution does not end the need for strong CA systems based on DRMs, but it does change their role. The emphasis becomes on keeping the great majority of people honest, when in the absence of any content protection they might be tempted to bypass legitimate channels to get content for free. The CA also continues to enable operators to detect who is viewing their content so that they can charge for it. Given this shift it might appear that the CA no longer needs to be as strong, but either way the future lies with ASIC core based CAs in conjunction with forensic watermarking for both revenue protection and to combat content redistribution.