dw2

30 June 2015

Securing software updates

Software frequently goes wrong. That’s a fact of life whose importance is growing – becoming, so to speak, a larger fact of life. That’s for three reasons:

  1. Complex software is spreading more widely into items where, previously, it was present (if at all) only in simpler form. This includes clothing (“wearable computing”), healthcare accessories, “connected home” consumer goods, automobiles (“connected vehicles”), and numerous “Internet of Things” sensors and actuators. More software means a greater likelihood of software error – and a greater likelihood of being hacked (compromised).
  2. Software in these items is increasingly networked together, so that defects in one piece of software can have effects that ricochet unexpectedly. For example, a hacked thermostat can end up reporting industrial secrets to eavesdroppers on the other side of the planet.
  3. By design, modern-day software is frequently open – meaning that its functionality can be configured and extended by other pieces of software that plug into it. Openness provides the possibility for positive innovation, in the way that apps enhance smartphones, or new themes enhance a webpage design. But that same openness enables negative innovation, in which plug-ins subvert the core product. This type of problem arises due to flaws in the set of permissions that expose software functionality from one module to another.

All three of these factors – the intrinsic defects in software, defects in its network connectivity, and defects in permission systems – can be exploited by writers of malware. Worryingly, there’s a mushrooming cybercrime industry that creates, modifies, and deploys increasingly sophisticated malware. There can be rich pickings in this industry. The denizens of Cybercrime Inc. can turn the principles of software and automation to their advantage, resulting in mass-scale deployment of their latest schemes for deception, intrusion, subterfuge, and extortion.

I recently raised these issues in my article “Eating the world: the growing importance of software security”. In that article, I predicted an imminent sea-change in the attitude which users tend to display towards the possibility of software security vulnerabilities. The attitude will change from complacency into purposeful alarm. Companies which are slow to respond to this change in attitude will find their products discarded by users – regardless of how many “cool” features they contain. Security is going to trump functionality, in a way it hasn’t done previously.

One company that has long been aware of this trend is Redbend (which was acquired by HARMAN in summer 2015). They’ve been thinking hard for more than a dozen years about the dynamics of OTA (over the air, i.e. wireless) software updates. Software updates are as much of a fact of life as software bugs – in fact, more so. Updates deliver fixes to bugs in previous versions; they also roll out new functionality. A good architecture for efficient, robust, secure software updates is, therefore, a key differentiator:

  • The efficiency of an update means that it happens quickly, with minimal data costs, and minimal time inconvenience to users
  • The robustness of an update means that, even if the update were to be interrupted partway through, the device will remain in a usable state
  • The security of an update means that it will reliably deliver software that is valid and authentic, rather than some “Trojan horse” malware masquerading as bona-fide.

According to my email archives, my first meeting with representatives of Redbend was as long ago as December 2002. At that time, I was Executive VP at Symbian with responsibility for Partnering. Since Redbend was one of the new “Platinum Partners” of Symbian, I took the time to learn more about their capabilities.

One person I met in these initial meetings was Gil Cordova, at that time Director of Strategic Marketing at Redbend. Gil wrote to me afterwards, confirming our common view as to what lay ahead in the future:

Redbend deals with an enabling technology and solution for OTA updating of mobile devices.

Our solution enables device manufacturers and operators to update any part of the device software including OS, middleware systems and applications.

The solution is based on our patented technology for creating delta-updates which minimize the update package size ensuring it can be cost-effectively sent and stored on the device with little bandwidth and memory consumption. In addition we enable the update to occur within the device memory constraints ensuring no cost-prohibitive memory needs to be added…

OTA updates can help answer the needs of remote software repair and fixing to the device software, as well as streamline logistics when deploying devices…

At that time, some dozen years ago, the idea that mobile phones would have more and more software in them was still relatively new – and was far from being widely accepted as a good thing. But Redbend and Symbian foresaw the consequences, as in the final paragraph of Gil’s email to me:

All the above points to the fact that if software is a new paradigm in the industry then OTA updating is a very crucial and strategic issue that must be taken into account.

OTA has, indeed, been an important issue since that time. But it’s my view that the full significance is only now becoming apparent. As security is poised to “eat the world”, efficient and reliable OTA capabilities will grow yet further in importance. It will be something that more and more companies will need to include at the heart of their own product offerings. The world will insist on it.

A few days ago, I took a closer look at recent news from HARMAN connected services – in particular at its architecture for cybersecurity. I saw a great deal that I liked:

Secure Car

  • Domain isolation – to provide a strict separation between different subsystems (e.g. parts of the overall software system on a car), with the subsystems potentially running different operating systems
  • Type-1 hypervisor – to isolate different subsystems from hardware resources, except when such access is explicitly designed
  • Driver virtualization – to allow additional peripherals (such as Wi-Fi, cameras, Bluetooth, and GPS) to be added quickly into an existing device with the same secure architecture
  • Software update systems – to enable separate remote software management for the head (dashboard) unit, telematics (black-box) unit, and numerous ECUs (engine control units) – with a 100% success record in deploying updates on more than one million vehicles
  • State of the art FIPS (Federal Information Processing Standard) encryption – applied to the entirety of the update process
  • Intrusion Detection and Prevention systems – to identify and report any malicious or erroneous network activity, and to handle the risks arising before the car or any of its components suffers any ill-effect.

I know from my own background in designing software systems that this kind of all-points-considered security cannot be tacked onto an existing system. Provision for it needs to be designed in from the beginning. That’s where Redbend’s long heritage in this space shows its value.

The full benefit of taking an architectural approach to secure software updates – as opposed to trying to fashion security on top of fundamentally insecure components – is that the same architecture is capable of re-use in different domains. It’s therefore no surprise that Redbend software management solutions are available, not only for connected cars, but also for wearable computers, connected homes, and machine-to-machine (M2M) devices.

Of course, despite all these precautions, I expect the security arms race to continue. Software will continue to have bugs, and the cybercrime industry will continue to find ingenious ways to exploit these bugs. The weakest part of any security system, indeed, is frequently the humans involved, who can fall victim to social engineering. In turn, providers of security software are seeking to improve the usability of their systems, to reduce both the likelihood and the impact of human operator error.

This race probably has many laps to run, with new surprises ahead on each lap. To keep ahead, we need allies and partners who constantly look ahead, straining to discern the forthcoming new battlegrounds, and to prepare new defences in sufficient time. But we also need to avail ourselves of the best present tools, so that our businesses have the best chance of avoiding being eaten in the meantime. Figuring out which security tools really are best in class is fast becoming a vital core competency for people in ever-growing numbers of industries.

Footnote: I was inspired to write this post after discussions with some industry colleagues involved in HARMAN’s Engineering a Connected Life program. The views and opinions expressed in this post are my own and don’t necessarily represent HARMAN’s positions, strategies or opinions.

Advertisements

1 Comment »

  1. […] This blog post from David Wood explains the growing importance of secure software updates.  Wood focusses on Redbend’s architecture for cybersecurity which has developed from their origins as a leading provider of mobile phone FOTA update mechanisms.  A key observation is that the system architecture of such a solution, if designed right, has broad applicability.  It’s a lesson that Redbend are fully exploiting right now in respect of adjacent verticals: […]

    Pingback by Week 27 | import digest — 7 July 2015 @ 2:37 pm


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Create a free website or blog at WordPress.com.

%d bloggers like this: