Particle Subscribe latency for earthquake warnings


#1

Hi,

We have over the last few years built an earthquake early warning system in Mexico and Chile, which we are soon releasing as open source - data, hardware, software…

I am currently developing a very simple Iot alarm device that will complement our apps and allow people to receive the notifications seconds before an earthquake. We are keen to use the P1 due to its cost, but unsure if we go with a direct MQTT connection to our AWS IoT broker, or if we use the Particle API to pub/sub from AWS. The problem with the MQTT connection is the P1 is very memory constrained, and is causing issues (we only got it to work with Google Iot so far). If we use particle subscribe that could solve our issues but we are left with 2 concerns;

  1. will the particle cloud add significant latency (seconds matter…) ? could the particle team potentially help me here ? I wonder for example if we could move our detection system to the same provider and datacenter as Particle… ?

  2. what is the uptime of the particle cloud ? I am concerned rerouting may introduce a point of failiure…

Any thoughts very welcome !


Ten big global challenges technology could solve - can you help?
#2

There are two different ways you can get the events from the Particle cloud:

  • Webhooks allow a connection to be made to a server you control when an event is published from a device
  • Using the Server Sent Events stream, you make a TLS/SSL encrypted connection to the Particle cloud and keep the connection open. Events are delivered in real-time as they are sent.

SSE has the lowest latency but the keep-alive pings on SSE are only around once a minute, so it’s hard to detect when a connection stops working for critical applications. Using multiple SSE streams from different locations can help, as can using a combination of SSE and webhooks.

Webhooks and SSE streams originate from AWS East at this time, but this may change in the future.

For larger products, it is possible to get a service level agreement (SLA) for uptime.


#3

Thanks @rickkas7 ! SSE stream sounds great, especially if we are nearby in AWS East (I’ll need to check which we use). Who can I talk to regarding the SLA ?


#4

wow, nice one!
I was just reading about the ten big global challenges technology could solve and your project can potentially help address one.


#5

Thanks! we definitely hope that we can do some good. There are others working on similar solutions but we have the advantage of having collected years of raw accelerometer data in Mexico and Chile which we are using to train our algorithms and improve detection.

In April we will be releasing all our sensor data, software and sensor schematics so anyone can contribute and even build parallel networks ! if anyone is interested please sign up !!


#6

IMPRESSIVE!!

I already did, how can one pass on this?


#7

Thanks ! For now just the link https://grillo.io/openeew/ . In April AWS will be hosting this as part of their opendata program, and then we will do a general release. Our sensor schematics will follow a few weeks later :slight_smile: