Comment by mdorazio | Hacker News Reader

Comment by mdorazio | Hacker News Reader

Parent Full thread

Be careful with expectations of "possible" vs. "practical". 5G for IoT devices is a big deal in remote field and some industrial settings where wi-fi is not an option. Two examples for you:

I worked on a pilot project with a state government group that wanted remote monitoring of water levels in wetlands and ponds spread across a few hundred square miles of mostly-rural area. wi-fi obviously wasn't an option and getting a WAN mesh setup was a huge pain in the ass and had its own limitations. We ended up testing with 3G, switching on twice a day to report, but the power consumption and data costs were too prohibitive to be practical long-term. 5G power use per bit is supposedly going to be far lower than 3G/4G. Jury is still out on data cost.

I now work in the automotive industry where vehicle logistics can be an interesting challenge. Parking lots for completed vehicles awaiting transport outside factories can be spread across a square mile of land (more if remote overflow lots are used), with multiple locations. There are multiple use cases for inspecting, tagging, remote driving, etc. these vehicles using aerial or ground-based drones and teleoperation to significantly cut down on costs. However, those use cases depend on high bandwidth, low-latency connections and extending effective wi-fi to lots would be massively expensive. 5G is a pretty interesting option here.

As others have mentioned, vehicle-centric use cases abound, but there are some additional ones for industries like oil & gas as well.

5G power use per bit is supposedly going to be far lower than 3G/4G.

For shorter ranges because higher frequency.

For average per-bit cost, because more bits/sec in the higher frequency encoding.

But, for IoT there is the burden of turning the RF segment on, finding lock, sending. So, for small 'grams' of message it won't necessarily be as good as the per-bit cost for streaming data or long lived associations. And, shorter ranges: You will be deploying more base stations. Albiet with lower power budget too, but .. Nothing is free.

Is it realistic to expect 5G coverage in rural wetlands?

From what I understand, the range of a single 5G base station is significantly lower than 2/3/4G so they are having to install new cell sites in major towns and cities just for 5G coverage.

There are technologies belonging to the 5G family that target specifically IoT. They can coexist with “normal” 5G /4G in the same base station. These technologies have way more robust transmission schemes that make them operational even with 20db (100 times) worse SNR than normal mobile communications technologies like LTE. The flagship technology in this space proposed by 3GPP is NB-IoT.

Millimeter wave frequencies are just one aspect of "5G", the standard of 3GPP Release 15 (and newer) are improvements and build off the existing standards of LTE (releases 8-14). Its quite possible, and likely that "5G" will be deployed across all existing spectrum, including lower frequency. Much like LTE has been and continues to be deployed across existing spectrum that was 2G/3G.

The focus on millimeter wave is large part due to existing capacity problems in high density areas, or for high bandwidth applications.

Probably since electronics has become so miniaturized and cheap, thinking of one cell per building or even one cell per light pole, wouldn't be that odd.

Seems to me the 'lots of vehicles' communication can be solved using a mesh network with the vehicles as nodes. The trend is in that direction already and can serve a pre-delivery function.

What about LoRaWAN?

This is opportunistic and free for everyone to use. For a big number of use cases you need licensed spectrum to ensure some minimum standards of communication. Imagine what happens when a lot of WiFi access points are in the same small neighborhood.

> remote monitoring of water levels in wetlands and ponds spread across a few hundred square miles of mostly-rural area.

But wouldn’t it work for this use case he had?

It is possible. Unlicensed technologies have their advantages as well. You should check how the trade offs of each solution manifest on your use case and pick the proper one. For example in some cases, Lowrawan has better energy consumption, thus the battery of the sensor could last longer.

I believe that in the long term and as the number of devices per unit of area increases, licensed technologies will be the only viable option.

LoRaWAN is protocol build for LoRa physical layer, and it is opportunistic. Because of the fact that you can't have every message ACKed makes it unreliable in my opinion (workout minimum standards).

However I've seen that there is proprietary protocol symphonylink from LinkLabs, that uses LoRa. They claim that it allows for having every frame ACKed, even bulk data uplink (OTA firmware upgrades). If so, I would tend to pick such technology for pond water telemetry.

Isn't lora very low speed?

It is, but if you need to send water level twice a day, then you don't need much speed.

That's true but most of the examples in this article don't play to the strength of lora (extremely long range) and seem to involve its weakness (speed)

For the case of wet land level measurement, wouldnt periodic sms be sufficient?

A solar powered system out in the open should, theoretically solve the power issue.

What about tethered balloon base stations?

Reply on news.ycombinator.com