The (Al)lure of Ad Hoc Networks

Last week Amazon opened up its Sidewalk Network to third party developers. We find Sidewalk, and its class of networks, incredibly fascinating. They hold so much potential, to supply an important foundation for communications for the trillions of objects in the world, but that potential always seems to be just over the horizon, out of reach, arriving some day, but not quite today.

Sidewalk is an ad hoc network, meaning that is comprised of devices out in the world, and while they may be centrally coordinated there is no central planning in their deployment. Connectivity happens opportunistically, whenever there is a connection node nearby. In the case of Sidewalk, this means all the Amazon hardware devices – Ring doorbells, Alexa speakers and TVs or whatever – can connect to whatever is allowed to access the network. These nodes all have Day Jobs running their intended purpose, but they have a little side hustle providing connectivity to other ambient devices. So the ‘opening’ of Sidewalk means that other people can connect to the Sidewalk network now. This network is ad hoc because Amazon has very little control over where their devices end up, and third party devices can only connect to them when they are in Bluetooth (or Wi-Fi?) range of the Sidewalk-enabled nodes.Apparently, Amazon has been adding extra connectivity to their devices. Specifically many of them have LoRa chips (another topic we need to revisit soon) which provide long-ish range to backhaul the Sidewalk traffic from other devices.

The big appeal of ad hoc networks is that in theory they are very inexpensive to deploy. Done well, they ride natural communications cycles of chips that someone else paid for to perform other tasks. No need to build massive cell towers with expensive base stations. A few dollars of extra silicon is all it takes, multiplied by millions of devices. Of course, the flip side of this is that the availability of these networks is never reliable. In order to be viable, an ad hoc network needs massive scale. Take a look at the Sidewalk coverage map and it looks fairly widespread, touching all the major population centers in the US. However, zoom in a bit and it becomes clear that there are large pockets of urban areas that are barely covered.

If this were a cell phone network, users would not tolerate this level of coverage. However, that is not necessarily a non-starter for ad hoc networks. Not every use case requires 100% 24/7 connectivity. If you just need a piece of data once a day from some remote object, the statistics behind ad hoc networks probably work. And there are a lot of situations where this works just fine.

That being said, none of the attempts to build wide area ad hoc networks have really taken off yet. We think there are two reasons for this. First, no one, not even Amazon, have a big enough network yet. All these networks probably need to be an order of magnitude or two bigger. Sidewalk likely has single digit million nodes, maybe into the low teens, but those coverage gaps are still pretty large. This will take time to bridge. Amazon has subsidized Sidewalk deployments largely by bearing the added cost of those connectivity chips in their Alexa devices, which are not that profitable to begin with. There are some companies working on clever ways to expand this further, but, it is going to take a few years for someone to stitch together enough nodes to make these commercially viable.

The second problem is trickier – who are the customers? Anyone looking for connectivity solutions already has several options. If these are too expensive, then those companies are likely to just limit connectivity. With cellular connections already incredibly cheap, the demand for ad hoc networks is too just too small for most companies to hang their revenue expectations on just yet. Of course, this is a chicken and egg problem. No one wants to sign up for an anemic network, but no one will build the network until they can hope to find paying customers.

That being said, we think this is an immense opportunity for someone, if they build it the right way. The cost of device-level connectivity continues to fall dramatically. Privately-held Wlliot is making printable bluetooth tag which sell for less than a dollar. They are going to need networks to connect those chips to. Wiliot is trying to build that network themselves, but if even Amazon is struggling to gain scale in this, it should be clear that Wiliot will not be the only ones to build one of these networks. Done well, there is a huge network waiting to be built.

Leave a Reply