Particle Mesh update — a note from the CEO

Originally published at: https://blog.particle.io/2020/01/28/mesh-deprecation/

Particle is discontinuing development of Particle Mesh, our OpenThread-based mesh networking solution, and will no longer be manufacturing the associated Xenon development board. Particle will continue investing in its other “Gen 3” products, the flagship cellular (Boron) and Wi-Fi (Argon) product lines. Read more below on the reasons why and the upcoming deprecation timeline, or watch a recording of our live stream Q&A below.

Particle’s mission is to make it possible for anyone from an individual developer to a large-scale enterprise to develop and bring to market an IoT product by wrangling the complexity of connectivity. With the Spark Core, Photon, and Argon, your IoT products could connect to the cloud via Wi-Fi in minutes. With the Electron and Boron, cellular connectivity for IoT no longer required deep networking expertise and a rolodex of connections at major telcos. When we added 802.15.4 mesh networking to our roster as the next layer of connectivity for IoT, we believed the technology and timing was right.

After one year in the field, we are making the difficult decision to end the development of Particle Mesh. We know the investment we asked you to make and we are inspired by what you’ve built with Particle Mesh. Unfortunately, the decision boiled down to weighing the major investments necessary to make mesh networking totally seamless and invisible against the realization that 802.15.4 mesh networking turns out to not be the right solution for most of the customers who wanted to use it.

This decision doesn’t affect our commitment to our “Gen 3” products, which are powered by a Nordic nRF52840 chip in a Feather form factor. We will continue to develop and support our flagship cellular (Boron) and Wi-Fi (Argon) product lines, and will continue to invest in our Bluetooth Low Energy support for local communications between devices.

For those that have purchased Xenons, our only Mesh-specific product, we will be sending you a store credit; watch for an email in the next few days. If you purchased through a third-party distributor, please see here.

Particle Device Cloud will continue to support setting up new Particle Mesh devices through the end of 2020, and the 1.6.x release of Device OS in March will be the last release to include support for Particle Mesh. Continue reading below for more details on the deprecation roadmap, ways to continue your own exploration of mesh networking, and context on how we arrived at this decision. And join me on a livestream Q&A this Thursday (1/30/2020) where I’ll be answering your questions.

Why we decided mesh networking is not for Particle today

When we began development on Particle Mesh in late 2017, we were beginning to see that our customers’ IoT products don’t exist in a vacuum. More and more often, our customers were looking to connect their products not just to the cloud via Wi-Fi and cellular connections, but to other nearby devices — typically to build out low-power wireless sensor networks around their core products. We saw that, by doing so, those customers were taking on a big new technical challenge; they were getting into the business of managing remote wireless networks without any good tools available. That looked like a great opportunity for us to help.

However, while that opportunity is real, we ran into two barriers that keep us from delivering on our promise with Particle Mesh. They are:

  1. Mesh networking, while a compelling technology, is extremely complex, and trying to make it “just work” with zero configuration for all customers in all environments just wasn’t feasible. As a result, the experience of Particle Mesh doesn’t live up to the standard that we set for ourselves with our other products.
  2. It turns out that, for most of our customers, 802.15.4 mesh networking seems to not have been the right technology in the first place; most customers would be better off using Bluetooth Low Energy (which we will continue to invest in) or a sub-GHz radio like LoRa (which can be added to Particle through third-party accessories).

Let’s explore each of those in a bit more detail.

1. Mesh networking is complex and hard to abstract.

Before we launched our Wi-Fi and cellular product lines in 2013 and 2015, respectively, those technologies were very inaccessible for IoT. However, they were already in widespread use for laptops and smartphones. We didn’t need to invent any new networking technologies to make it possible to build a Wi-Fi or cellular connected IoT product; we just needed to redesign the whole experience around IoT. Our goal was to make it so that you didn’t need to have any deep expertise in wireless networking to be able to build an IoT product; the networking layer should be abstracted so that your product “just works.”

Our goal with Particle Mesh was the same. Particle Mesh is built on top of OpenThread, the only IPv6, open source, IoT-friendly mesh networking protocol available – and we saw an opportunity to create an “it just works” experience on top of it. OpenThread was originally developed to support Google Nest smart home products, and had recently become available for companies outside of the walls of Google Nest. The Thread Group was interested in seeing OpenThread used outside of the smart home and in commercial contexts, where most of our customers operate.

Illustration credit: the Thread Group.

But since we were the first ones taking OpenThread in that direction, and because it was and still is a new technology, we discovered the ways in which it wasn’t ready for us. There were three key areas where the current state of mesh networking technology presented hurdles: manufacturing at scale, deployment at scale, and reliability at scale. Critical requirements like configurability on the manufacturing line, roaming between networks, and expected power budgeting were just not available in the current iteration of the mesh protocol.

In addition, we came to realize that building a proper network management solution that would meet our expectations for simplicity while being powerful enough to support large-scale deployments would be a massive undertaking, deserving of an entire company on its own. As we made plans for our 2020 product roadmap, we couldn’t justify dedicating all of our engineering resources toward a massive “moonshot” that is still exploratory in nature at the expense of our core platform for cellular and Wi-Fi connected products.

2. 802.15.4 mesh networking isn’t right for many customers.

It’s 2020, and you’re looking to build a local wireless network of IoT devices. What are your choices? You’ll probably consider the following three paths:

a.) Use Bluetooth Low Energy to connect a handful of wireless sensors within a range of 100-300 feet.

b.) Use LoRa or another sub-GHz radio to connect wireless sensors that are further apart or are separated by obstacles and barriers (walls, floors, etc).

c.) Use 802.15.4 mesh networking (primarily Thread, but also Zigbee and other protocols) to construct a large-scale wireless network with many nodes that can rebroadcast messages across the network.

We were excited about Thread-based 802.15.4 mesh networking because of its inherent resilience – the fact that, in a well-functioning network, if any individual node goes down, the network self-heals. We care a lot about reliability and that seemed like a killer feature to us.

But in practice what we found is that our customers would be better served by BLE or LoRa. Most of our customers are using a very small number of nodes in a network, which means that BLE is sufficient (and much less complex). And other customers are operating in environments like factories or basements where 2.4GHz radios just don’t have the necessary range and LoRa would be a better fit. For example, one customer wanted to implement mesh in industrial refrigeration units. The 2.4GHz radios don’t have long enough range or the right physical characteristics to transmit through the refrigerator wall. While the intention of Particle Mesh was to be a solution for industrial use cases that needed resilient, device-to-device communications, in the end it could not meet the needs for our customers.

We took a risk, made a bet, and brought you along for the journey – and in the end we did not deliver what we intended. Now our goal is to make it right, and to take our learnings from the past year to guide you down the best path for your IoT product or project.

Particle Mesh Deprecation Timeline

Xenon devices and Particle Mesh networks that have already been setup or that are setup before the end of year 2020 will continue to function normally with access to the Device Cloud and our existing device management features. The timeline is as follows:

  • March 2020: Device OS release of 1.6 will be the final release to include Particle Mesh support
  • June 30, 2020: Technical support via the official support portal for Particle Mesh devices will end; customers may reference community support in the forums
  • Dec 31, 2020: ability to set up new mesh networks, or add devices to existing mesh networks will be removed from the mobile app and cloud APIs
  • Jan 31, 2021: Network management features (the “networks” page) in the Console will be removed

If you would like to continue exploring mesh networking with your Particle devices, we’ve outlined ways to do that below. The full roadmap can be found here for reference.

Other ways to use your Xenons

There are a few ways to continue exploring mesh networking with your Xenon development board as well as accessing OpenThread. You’ll find more documentation on how to do so here.

Use your Xenon for Bluetooth Low Energy: You can continue to develop with Particle’s Bluetooth Low Energy APIs as long as you activate your Xenons before the end of 2020. Keep in mind the last Device OS to support Particle Mesh will be released in March, so these devices cannot be upgraded beyond that version. You can also use the Xenon disconnected from the Particle Device Cloud (manual mode) to act as a BLE end point, though it must be flashed via Workbench or the CLI tool and does not include OTA capabilities. More information here.

Develop on OpenThread SDKs directly: You can access the Nordic nRF52840 SDKs to develop with OpenThread’s mesh networking protocols. This will require a third-party gateway and will overwrite Particle’s Device OS. More details here.

Reprogram the Xenon with Circuit Python and Adafruit: You can continue to use the Xenon as an IoT development board by reprogramming it with Circuit Python and working with Adafruit’s other Feather-compatible products — instructions here. This will overwrite Particle’s Device OS.

Looking forward to what’s next

Deciding to end development and support of Particle Mesh was a difficult, painful decision. Our mission is to reduce the barriers of developing and bringing to market IoT products, and we have failed in that mission here; we have given you a path and allowed that path to end.

But the silver lining here is that this decision frees us up to invest more deeply in the technology that we’re most excited about – cellular connectivity for IoT.

We are strong believers in the future of cellular IoT. Since the launch of the Electron in 2015, more and more of our customers have been using our cellular connectivity platform to do everything from monitoring heavy equipment to tracking the location of electric bikes and scooters. Cellular is great in large part because “it just works;” if your device is near a cellular tower, it connects out of the box. That’s it. Yes, developing a cellular product is complicated by telco relationships and nasty certifications and by a complex and esoteric networking stack – but that’s where we come in.

We’ve already made it possible to build 2G, 3G, and LTE CAT-M1 connected products with Particle. We’re now working on support for additional LTE radios to provide global LTE coverage and better performance. We’re doubling down on our most powerful features – over-the-air firmware updates and device diagnostics – and we’re working on new solutions specifically for mobility and asset tracking, which have become some of the most common use cases for Particle.

There are more great things to come from Particle, and we look forward to sharing them with you soon. Stay tuned!

15 Likes

I appreciate the clear message Zach. For those of us who want to view the livestream but can’t due to other commitments, will there be a recording available to those of us who register for the event?

Yes, a recording will be made available.

3 Likes

That’s good to hear as I’m getting an error trying to sign up.

Do you have an adblocker running? This sometimes causes issues with the form.

2 Likes

:sweat: Yep. That fixed it. Usually I’m better about remembering that they can cause issues.

I guess I should be saving this for Thursday, but what about Ethernet Featherwing? Does it have a use besides as a backhaul for a Xenon? Will promotional credit be provided for it?

1 Like

Very very disappointing for anyone using xenons in a commercial product already.

7 Likes

The EtherWing (as I like to call it) can also be used with the Argon and Boron too.

Have some mesh products still in the packaging. Can I get a refund?

2 Likes

I am also really disappointed in this. I just spent the last 2 weeks developing a product that I was really excited about. The whole “cellular” just works isn’t true. I don’t want depend on an internet connection to pass data between devices.

3 Likes

Your lucky only spending 2 weeks. We’ve just spend 6-8months developing a couple of products which we will now have to ditch.

Oh well it’s only money!

6 Likes

yeah 2 weeks is kinda short i agree but I just got everyone all excited about it. I have a working prototype and all the nighter ups were impressed. Now I am going to look like an ass.

1 Like

I was an early adopter of your mesh boards. This announcement is me done with particle for purchases or subscriptions. I’ll use the cloud services for things I already own, but you’re not getting one cent more out of me. Just be honest; subscription cellular is the moneymaker and you have no interest in supporting one off purchases like the xenon.

2 Likes

Also would like to chime in and say thanks as I’ve been building an entire product dependent on the mesh aspect of the gen 3 devices.

This will definitely hurt the startup that I’m doing this for.

I wish you would just take a step back and rethink the mesh approach (and maybe just start over on it) rather than just ditch mesh entirely…the hardware is there, maybe the firmware approach just wasn’t.

If you build a stable and easy to use mesh feature (I don’t think think the previous was completely either), the market will be there.

3 Likes

1 year and 2 months here, with investors lined up. I hope I can find a replacement soon.

2 Likes

the funny thing is I would have bought another $100 worth of devices today if the Xenons were not out of stock. Now I will look for something else. My trust in Particle is pretty much gone.

6 Likes

Agreed. This is a Google move. Kill something then expect everyone to want to use the rest of the products because they’re not going anywhere, promise!

4 Likes

Sorry to hear that your project has been disrupted. Hopefully you’re able to find a good path forward with OpenThread or CircuitPython, or by using a LoRa solution if that’s a better fit.