For the past month I’ve been in Seattle, taking a break from the Austin heat. One of the things I’ve done while here has been to meet with M87, a startup that actually came to life in Austin five years ago before getting funding from Madrona Ventures and moving to the Pacific Northwest. The company makes software that runs on wireless devices and can boost the signal performance of all devices in a selected network.
The software runs on the devices themselves to generate what becomes a self-optimizing network that can use the processing power and connectivity of nearby devices to boost the overall network capabilities of all the devices in the network. The software was designed to run on smartphones, and includes policies that dictate how devices borrow connectivity from one another.
These policies may include not borrowing processing power from devices that have little battery life left, not using capacity from phones on a 3G or 2G network, or avoiding those that are roaming. Cellular providers are currently the target client for the software; M87’s software helps them boost network capacity by using customers’ existing computing power on their phones.
But M87 realized it also could use this ability for other networks and devices, not just cellular networks and smartphones. For example, it has a few trials in manufacturing plants on alternative networks such as Wi-Fi and Bluetooth. The current focus in on helping companies create higher-performing and more reliable wireless networks through use of the self-optimizing network technology.
Wireless networks in industrial settings can be tough to operate. There are often large amounts of moving metal or even liquids around, which can wreak havoc on signal quality. Or the overall area can be huge and therefore tough to adequately cover. Add to this the fact that many machines need both reliable coverage and low-latency coverage and wireless becomes a non-starter.
So M87’s networking boost could be a compelling option for companies trying to connect factories without all of the expense of wires. But I think the distributed policy engine is actually as interesting if not more interesting for industrial IoT deployments.
In the cellular world, carriers are running about 30 different policies on the distributed smartphone network. If you think about industrial IoT, having the ability to enforce policies at the edge on a self-optimizing network could be useful in giving machines more autonomy. Instead of prioritizing for battery life on a cellphone, policies in industrial IoT settings could send certain types of data faster than others. For example, an anomaly that could affect worker safety might get enough compute and wireless capacity to ensure someone doesn’t get hurt.
This could be really interesting for manufacturing or robotics. Current versions of the software can run on Linux-based devices and require about 1 MB of memory. So it isn’t going to boost capacity for sensors anytime soon, but it could easily work for industrial process controllers or gateways.
I haven’t seen a lot of startups that can provide a way to run a set of governing policies in a distributed fashion on unreliable networks. But M87, through the use of distributed computing, can, all while boosting the capability of the network to a point where it might solve some of the bigger problems at the edge.
I look forward to seeing how the few tests it has running pan out.