thelittleshort
Top Bloke
What about XMOS?
Article below by their CEO Mark Lippett in Forbes. In my opinion Mark is basically describing the problems he knows he can fix using Akida IP - due to the partnership between XMOS and Plumerai
BrainChip smart homes and cybersecurity
INNOVATION
Privacy, Intelligence, Agency: Security In The Smart Home
Mark Lippett
Forbes Councils Member
Forbes Technology CouncilCOUNCIL POST| Membership (fee-based)
May 5, 2022,06:30am EDT
CEO of AIoT chip company XMOS, Mark Lippett is a technology leader with 25 years’ experience in startup, scale-up and blue-chip companies.
While the dawn of the smart home has been heralded for years, the establishment of smart speaker technology means we’re finally actually entering the era of intelligent, responsive homes.
Home automation is projected to expand its value to $75.3 billion by 2025, with a compound annual growth rate (CAGR) of just under 16%. Smart speakers are the key driver of growth here. Data from Statista suggests that 320 million smart speakers were deployed as of 2020, projected to double by 2024.
With pandemic-related factors accelerating this growth as people spend more time indoors, concerns over security and privacy come to the fore. Many IoT environments are failing to properly protect their users, given their over-reliance on cloud infrastructures and internet connectivity.
Balancing Innovation With Privacy
Having Amazon Alexa or Google Assistant may be convenient when turning off the lights or ordering in, but it also offers an open window into your home with the possibility of thousands looking in, listening from anywhere in the world.
Cloud connectivity and the “always-on” power model raise significant ethical concerns, with consumers understandably wary of their every word being recorded and transferred online.
Despite the short-term financial incentive, this model may be self-defeating. Bringing smart home architectures to market will be harder if privacy concerns aren’t addressed—and that’s true not only for today’s devices but future models, too.
Manufacturers need to strike a technical balance between innovation and privacy for our smart homes to evolve. Devices need to interpret user activity and context and respond—that’s their purpose. They’re not “smart” if they can’t perform this function.
However, residents must also feel secure in their own homes: recognized, not watched. Striking this balance will be crucial for device engineers going forward.
Process Your Data Locally
The key to providing secure, private and responsive homes is reducing dependency on the cloud, instead embedding intelligence within the home environment itself.
At present, our devices can’t process user inputs alone. Sensor data must be transferred to the cloud for interpretation and contextualization before instructions can be carried out. This causes latency issues where smart homes take longer to respond. At best, this is a pain; at worst, tasks or value-adding use cases are rendered useless due to resulting health and safety concerns.
Thankfully, we have an alternative to the cloud: the artificial intelligence of things (AIoT). This model embeds intelligence and processing power directly into the smart home device, processing commands locally. I’ve written about it before for Forbes, examining the convergence of AI and IoT devices.
However, there’s a challenge in bringing this edge intelligence to market. AI chips are expensive, and it’s difficult to configure them for compatibility with home devices.
Making The AIoT Work
The chips that drive such home devices must deliver a combination of AI, DSP, control and communications. Economics demand that these are delivered in a single device, affording designers greater control over how these four attributes are balanced.
Manufacturers will also need to deliver this in smaller packages with low overall BOM costs. Creating these programmable, efficient devices with AI capabilities won’t be easy, but it’s an essential piece of the data protection puzzle.
Rather than rely on the cloud for context, AIoT-enabled chips would enable more sophisticated sensor processing—delivering face and image identification, presence detection and even life sign monitoring to capture rich, contextualized information and react independently.
Simply put, you won’t need to risk or surrender personal information to make the smart home work.
Security And Privacy Aren’t Negotiable
Realizing edge AI for smart homes would mark significant progress for not only consumer privacy but overall home security. Improvements to latency, processing and more sophisticated multimodal sensors will all contribute to a better-protected home.
For example, edge intelligence will allow smart home devices to distinguish between household occupants with ease. Devices will be able to identify and ignore potentially dangerous commands from young children, contact emergency services in case of medical emergencies and even alert household members if an unrecognized individual has entered the building.
In the longer term, a symbiotic relationship would allow devices to “borrow” intelligence from one another to build sophisticated room-to-room continuity. Alarm clocks can initiate coffee machines; gas sensors could disperse dangerous gases by opening windows; and security lights in the back garden could lock the front door. It all depends on the sensors and intelligence available.
However, with consumer concerns around privacy a problem for right now, such devices need to balance perceptiveness with some tact—and that means the edge. In removing the reliance on data transference to the cloud, edge devices pose very little risk to consumers’ privacy. On-device processing allows smart home tech to dispense with data once it’s used. In many cases, little-to-no data would need to be stored in any format.
All of this drastically reduces the amount of data leaving the house, as well as the number of attack surfaces the smart home is exposed to. It means a more cybersecure home, one that’s better protected against data exploitation by private corporations.
We can only make that a reality by addressing the growing concerns over consumer privacy, with 63% of consumers considering connected devices “creepy“ in terms of data collection and behavior, while 50% of U.K. consumers were “fairly concerned” about the use of their data, according to a Deloitte study from 2020.
Choosing to make devices independently intelligent, rather than relying on the cloud with its associated data transfer, is a critical enabler for an exciting new smart age—free from invasion of privacy.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Follow me on LinkedIn. Check out my website.
Mark Lippett
CEO of AIoT chip company XMOS, Mark Lippett is a technology leader with 25 years’ experien…
Read More