Beyond IoT: Building Decentralized, Intelligent Infrastructure

As I wrote recently, the Internet of Things (IoT) has been experiencing, at a minimum, some serious growing pains.  This is particularly true for consumer IoT where a lot of old issues (interoperability) remain, while others (security) are becoming more concerning.  With a few bright exceptions, many consumer IoT products solve first-world problems, often representing a marginal improvement over existing solutions.

But the IoT was always meant to be more ambitious and exciting than just the smart home, the factory or other discreet “single-player mode” use cases.  The internet of things was always about networks, where connected objects could be tracked and activated across wide geographic areas, supply chains, health systems and other contexts representing trillions of dollars of economic value.

Rather than IoT,  perhaps we should start using the expression “intelligent infrastructure” more frequently to describe those networks.  With the parallel progress of machine learning at the edge, intelligent infrastructure will enable software-based intelligence to permeate the physical world, enabling real-time optimization and orchestration of connected “things” (objects, vehicles, machines, buildings), at a system level.  Uber, Lyft and others give us perhaps the closest approximation what such networks could look like at scale, except that, in an intelligent infrastructure paradigm, such communications would be machine-to-machine, with no human in the loop.

Building blocks: the connectivity issue

The problem is that, for all the noise about the Internet of Things over the last few years, we are just starting to deploy the first building blocks necessary for such intelligent infrastructure to become a reality.  As often in such circumstances, there is a chicken and egg problem. Infrastructure is necessary for applications to thrive, but developers of infrastructure want to see momentum in application development before doubling down on their investment.

Connectivity, in particular, is a surprisingly unsolved problem. Certainly, technologies such as Wi-Fi, Bluetooth, Ethernet, Zigbee and Z-wave are adequate for short-range connectivity.  But for the kind of network scale required for intelligent infrastructure, it is very much early days.

Connecting things over a long distance is an arduous task that has very specific requirements:  connectivity needs to be long-range, low-bandwidth, battery-efficient (as you won’t go around changing batteries in millions of connected objects) and cheap (many of those things will very small amounts of data, often quite infrequently).  Traditional long-range solutions such as cellular or satellite are way too expensive for this scenario.

Solutions have started appearing.  Some are startup efforts, such as Sigfox (a dedicated cellular IoT network) and LoRaWAN (a standard promoted through the LoRa alliance).  Those are Low-Power WAN (LPWAN) wireless technologies, designed specifically to interconnect low-bandwidth, battery-powered devices with low bit rates over long ranges.  In addition, over the last year or two, big carriers have jumped into the action, with two types of offering: Narrow-Band IoT (or NB-IoT, used by Deutsche Telekom, T-Mobile and Dish Networks) and LTE Cat M1 (Verizon and AT&T).  And there is a lot of hope (and hype) around 5G for the most intense IoT use cases (such as autonomous cars), but it is years away at best, at least in the US.

All the above are encouraging solutions, but there are still a lot of trade-offs involved.  Take cost for example — the cheapest NB-IoT offering will still cost $6 a year per connected device; the necessary hardware modules are still quite expensive, and building at the chip level requires tremendous firmware expertise as the cellular stack is very complex.

The Helium network

Is there a better way? I will speak about one example I am familiar with.  Over the last couple of weeks, Helium, a San Francisco based startup in which my firm FirstMark is a proud investor, alongside Khosla Ventures, GV and Munich Re, introduced a really interesting new network concept, which has the potential to be the best of all worlds and substantially accelerate the rise of intelligent infrastructure.

Helium is a connectivity network for machines – it will enable the long-range connection of devices to the Internet in a way that is cheap, secure, open and battery-efficient.

You can think of the Helium network as a two-sided marketplace, not entirely dissimilar to, say, Airbnb — with the important difference that Helium will not charge participants on each transaction (more on this later).  On one side of the transaction, you have the network participants that need the connectivity, for example, enterprises that need to connect machines to the Internet (in the Airbnb analogy, they would be the people renting the apartment).  On the other side of the transaction, you have participants that provide connectivity (in the Airbnb analogy, they would be the people owning the apartment).

How do they provide connectivity? They use physical devices called gateways, which they can either buy from Helium or build themselves, as the specs are completely open.  Gateways can be installed anywhere – in your company, store or home window.

Gateways help connect devices to the Internet using a new wireless protocol, called WHIP, which is developed by Helium, but completely open.  WHIP is a remarkable protocol in that it is both low-power and highly efficient. With WHIP, devices can communicate over many square miles in dense urban environments or hundreds of square miles in rural settings.  To do so, it doesn’t require any kind of special hardware (like GPUs), and it essentially consumes the same amount of power as an LED light bulb.

A decentralized approach

Now, why would anyone want go through the effort of becoming a connectivity provider in the Helium network? This is where the Helium blockchain and Helium token come in.

I know… it is difficult at this stage of the crypto hype to not roll one’s eyes at the mention of another blockchain solution.  

Also, the intersection of two buzzy terms such as IoT and blockchain seem to be the type of things venture capitalist dreams are made of, but an otherwise very perilous place to be.

Now, processing a massive amount of IoT transactions on the blockchain itself is challenging, due to scalability issues, as widely acknowledged by Vitalik Buterin and other luminaries. However, Helium is not trying to run the IoT on the blockchain, but “merely” connect machines to the Internet, so that data-intensive computations can be processed in the cloud.  

In Helium’s case, using a blockchain-based approach makes a lot of sense, and may in fact turn out over the next few months to be one of the most interesting use cases of blockchain technology out there.

I was privileged to start working with Helium almost from inception (I literally wrote the first non-founder $250,000 check in the company back in 2013), and the concept at the time was very much what it is today: build a broad IoT network, where connectivity would be provided not by installing massive cell towers, but through a distributed approach involving people and companies deploying small gateways that would eventually add up to create network density.  However, beyond concerns that new and cheap connectivity technologies would quickly appear given the size of the IoT market (this, interestingly, never happened), we could not convincingly figure out how we would incentivize people to join the network, quickly and in large numbers, except through paying them sums of money that would would add up and ruin the entire capital efficiency of the model. So Helium took a detour for a few years building and selling IoT connectivity technology in “single player mode” for Fortune 1000 companies.

Fast forward to today, and the emergence of the blockchain now makes possible what was not at the time.  The core idea of a token economy is precisely to enable the creation of networks and solve the “cold start” problem that plagues all new networks (nobody wants to join a network if there’s nobody relevant on it already).  By enabling participants to earn tokens by undertaking certain activities desirable for that specific network, blockchains help incentivize people to join networks early in their existence (while it’s still easy to obtain tokens, and the value of tokens is at the beginning of the appreciation curve), and align economically the various participants long term without requiring a trusted relationship between them, or the blessing of a central authority granting permissions or incentives on the network.

In Helium’s case, the desired activity is to provide wireless network coverage that things are able to use to connect to the Internet.  Therefore, participants that undertake that activity (“miners”, in crypto parlance) will be rewarded with tokens.  Users (machines) that use this wireless network coverage pay tokens to the miners to send data to and from the Internet.

Providing network coverage in the Helium network is a physical activity involving specific hardware, making Helium almost a “physical” blockchain, and this is arguably a new concept in the crypto world.  For comparison, in the bitcoin world, miners earn tokens (bitcoins) by solving complicated math puzzles, providing computational power that secures the network (what’s known as “Proof of Work”).

In contrast, in the Helium network, miners prove that they are providing (or verifying that other miners provide) wireless network coverage through a brand new mechanism called “Proof-of-Coverage”.  This mechanism establishes that miners are operating the right type of radio frequency hardware and firmware that’s compatible with WHIP. It also establishes that the miners are indeed established in the location they claim, therefore providing the coverage they claim to be providing.

Helium joins a small club of companies that are truly leveraging blockchain technology for its ability to accelerate the creation and administration of a decentralized network, such as Protocol Labs (IPFS document sharing network) or Golem (computing resources sharing network).

It is worth noting that the Helium network will truly be a “Decentralized Machine Network”.  In particular, there will be no toll of any kind levied by the network for participants to generate and utilize network coverage, which is a radical concept, particularly compared to coverage provided by telecom companies. Helium will make money by providing additional services and products around edge orchestration, reporting and wallet management.

Next steps

The Helium Decentralized Machine Network will officially launch in Q4 2018. There will be no flashy, celebrity-sponsored ICO at Helium, which views the use of tokens as a long term way of building the network, and will issue tokens in a way it hopes will set a new standard in token issuance from a regulatory and compliance perspective.

Beyond blockchain-based incentives, Helium will ensure initial network density through a number of partnerships with individuals, companies and municipalities.  If you are interested in joining this effort, you are encouraged to reach out to them!

You can learn about Helium in their white paper, on Telegram, in the MIT Technology Review or on Vinod Khosla’s blog.

Helium offers the promise of a great building block to facilitate the emergence of intelligent infrastructure.  I am hopeful there will be many others around key areas (security, scalability, data analysis, etc.) that will help the Internet of Things realize its true potential.


Leave a Reply

Your email address will not be published. Required fields are marked *