AchimKraus If they don’t want to use a “custom endpoint”, what endpoint do they use?
It’s a JSON feed over HTTPS, they call them endpoints. They don’t want to make a new JSON feed with only the minimal data the project needs for us.
AchimKraus And if your company don’t want to run a server, why then not use directly, what the customer’s endpoint offers?
That is what we were doing. We have a new vendor to collect real-time data and standardize it now. This issue is/was the size of the JSON data. It contains a lot more information than needed for the project. At the required rate of transmissions the LTE data costs add up to be quite significant.
AchimKraus What are the security considerations, e.g. TLS and the used authentication mode?
That’s a great question! Verizon doesn’t mention any network level encryption for the NIDD service. It’s something I’ll have to get more information about from them. I have considered implementing some form of E2E encryption for the messages between the devices and our servers, but I’m not that far along yet in development.
AchimKraus Why do you consider, that a service, which does some data translation, provides benefits in your case?
I imagine a scenario where you don’t have full control over a data source and you want to modify it before sending it to your end devices.
For example, had I been able to do that in my previous project, I could have pre-processed the vendor data to significantly shrink the size and convert JSON timestamp objects to epoch timestamps. That would have reduced complexity and processing time on the device side.
You could also send useful data back if the device is unreachable. I.e. Time of disconnect, last know location, etc.
AchimKraus Why do you want to use NIDD instead of IP?
The value of unlimited messages for a low fixed cost just seems to good to ignore! In order to achieve that low fixed cost, NIDD is the only option I have found in my region. More on that below.
AchimKraus And why does it seems, that is you, to want and not your customer nor company?
Well, that’s because the current company behind this idea has one member. Me! It’s not worth hiring more engineers until I’m confident this is service people would want.
In regards to the previous project I keep referencing, I (currently) work part time with a regional transit authority in the US. Budgets are tight, resources and time are limited. I am the one who had to find compromises to get the project into a stable state they’re happy with and not break the bank. The project is still adding new deployments, but development time is mainly limited to maintenance. I’m happy with the results, but not completely satisfied with the amount of remote troubleshooting available and data consumption.
I see this as the opportunity to create the solution I envision for the cellular hurdles I have faced. But I need to know if that solution is something potential customers want as much as I do.
AchimKraus In my experience, the “tons” are rather using IP than NIDD.
I’ll admit part of the value I see here is probably due to the state of NB-IoT in the US. This Verizon NIDD plan is the only advertised NB-IoT service I have found in the US with a price tag. AT&T has discontinued NB-IoT services and T-Mobile offers it but doesn’t advertise any pricing or details. I will reach out to T-Mobile once this project gets enough interest.
Maybe in other countries you can find similarly low cost NB-IoT plans that allow you to use IP, but I have not found that option in the market I currently operate.
AchimKraus If you need to convert a message every 30s, the most services, which doesn’t belong to the MNO, will need to finance that “work load”, so it will be hardly some cents in a month for more than 80000 messages and 350MB. The MNO may compensate that by the SIM costs, but a “pure iot service” will need the money to run.
In my mind, the amount of resources required to process HTTP(S) calls and do some minimal amount of data processing isn’t that huge. But like you say, it adds up and I could be underestimating it. Part of why I’m posting now is to help me figure out what users want in a cellular service and what it’s worth to them.
Here’s a scenario I would love to get your thoughts on:
Imagine your in the US or UK and you have some device in the field that uses 2MB of data per month. Looking at Verizon’s price list, it’s $1.50/device/month for 1MB and $6.00/device/month for 10MB.
Of course we won’t be able to charge at cost, so this service will be more than the base $0.75/device/month. Imagine it costs $2.00/device/month.
Would you or your company rather pay the $6.00/device/month for 10MB, spend the engineering time to reduce data usage to achieve $1.50/device/month, or spend $2.00/device/month for our NIDD service?
You would be going from CAT-M to NB-IoT. That means increased latency, but also increased power efficiency. Although, the efficiency gains are probably offset by the increased number of transmissions in this case. You would also have to add the library we create to your project. Ideally that library will be as transparent as possible.
Does this seem like a realistic scenario to you?
Thanks for taking a look at my plan, and let me know if you have more questions!
