Two babies


A few end-of-year thoughts on P2P and LPWAN’s:

At Haystack we’ve seen an uptick in the number of solutions/concepts that are looking for a decentralized approach to implementing a low power, wide area network. This is driven in large part by blockchain, which I’ve started to talk about here and here.

Blockchains are driving discussions about decentralization, but those conversations are surfacing some other realities about the nature of decentralized IoT networks.

Centralization, as applied to the IoT, can describe the physical network topology of an IoT network, or it can describe the governance (application layer) of the network. A “hub-and-spoke” physical topology as is common in cellular networks is one example of a centralized physical network with a centralized governance model.

Semi-decentralized networks might include decentralized blockchain-based applications — let’s take a supply-chain app like VeChain — that operate over internet links run by hub-and-spoke telco’s with a centralized physical and governance model.

A truly decentralized IoT network therefore describes a non-hub-and-spoke (e.g. non-cellular) physical topology as well as a decentralized governance model.

Putting aside governance for purposes of this post, what constitutes a decentralized physical topology in an IoT network? For starters, it means eliminating a “hub” or central server and implies some version of peer-to-peer networking. In theory, you and I could have a decentralized IoT network with just two endpoints — one for you and one for me — communicating on a peer-to-peer basis.

If we add a third endpoint, our options may expand from simple peer-to-peer messaging, to multi-hop messaging. In other words, if A, B, and C are arranged in a straight line where A is out of range of C, A could send message X to B, who in turn passes X over to C. We can even add a fourth endpoint (“D”) and extend the process further. At Haystack, we refer to this as multihop networking.

A decentralized physical network topology eliminates the “hub” and instead relies on endpoints to communicate on a peer-to-peer basis and in some cases “hopping” or “meshing” a message from its origin to its destination.

While “meshing” was at least theoretically implemented with certain IEEE 802.15.4 protocols like ZigBee, we have yet to see meshing with longer range, low power LPWAN radios like LoRa. Nor are we likely to see it given the overhead it requires and the strength of signal propagation seen by LPWAN radios. In reality, meshing for 2.4GHz protocols was more a marketing gimmick driven by poor signal propagation than an actual need for “meshing” — that is to say, if 2.4GHz could have done the job in getting messages directly from an endpoint to the gateway via better signal propagation, there would have been little need for meshing.

Adding P2P + LPWAN connectivity to Cheeseburger Compass

The Cheeseburger Compass (CC) is a simple example of a piece of geolocation-enabled gadgetry that operates without the internet, more or less. And in its own simple way shows us another path towards a decentralized IoT: it just tells you the closest place to get a burger.

Hypothetically, we could add a low power, wide area network radio like LoRa to the CC. And in doing so, every CC user can communicate with other CC users over ranges of .25 miles to 10 miles. All are mobile, and using the right networking protocol, even with GNSS/GPS on the mobile client, the battery can last years.

And still no internet.

Five Guys, In-N-Out, and White Castle could interact with these CC users offering promotions, internet free. Using a networking protocol like DASH7, promotions could be broadcasted to CC users or routed individually.

Using a networking protocol like DASH7 that supports P2P communication, CC clients could send messages to one another.

As long as the message size remains small — a text message, email message, GPS location coordinates, sensor data, etc. — a LPWAN-based system can operate without the need to connect to the broader internet.

What About DNS?

A LPWAN based, decentralized messaging system could route messages via one or more application layer protocols and using a networking protocol like DASH7 that supports multi hop, a decentralized equivalent of a Domain Name Service could exist either on “master nodes” or in a blockchain environment.

You say: but isn’t blockchain using the internet? Perhaps, but not necessarily. A truly decentralized blockchain network could (theoretically) operate on a purely wireless basis. Some companies like Blockstream are taking this to extremes using satellites but this is only one example.

Final Thoughts

A decentralized IoT, focused on short bursty messaging, is not too different from a decentralized internet that only processes SMTP messages.

In lieu of meshing, expect “repeaters” to play a larger role in decentralized LPWAN deployments.

The importance of peer-to-peer messaging, especially by battery powered endpoints, in a decentralized IoT is difficult to overstate.