back to article Meet the LPWAN clan: The Internet of Things' low power contenders

LPWAN - low power wide area network - is the proposed connectivity tech of choice for powering the Internet of Things - and it comes in many flavours. An IDTechEx Research report put it into perspective recently, when it predicted that there will be 2.7 billion LPWAN IoT connections by 2029. What we are already seeing is that …

  1. Anonymous Coward
    Anonymous Coward

    VHS and Betamax

    This IoT thing is really VHS and Betamax all over again, except they are plenty instead of only 2.

    Tons of different technologies, with a single one remaining at the end. Why on earth wouldn't those people agree on something more common like z-ware or Zigbee ???

    1. Adrian 4

      Re: VHS and Betamax

      Zware and zigbee are short-range technologies like bluetooth and wifi - they're suitable for home use with a local gateway. These other options work at much longer range - 10km is mentioned for sigfox. This makes them suitable for small remote sensors like water level (flood warning), etc.

      Certainly, home sensors could use short-range technologies and probably will. But there's a large class of devices that want low power, low data rate and long distance to the gateway.

  2. ForthIsNotDead

    Interesting article

    We're making great progress with LoRa, both LoRaWAN and direct (end-to-end) LoRa communications, using in-house designed and built telemetry units that are very small and can run for years on 4 AA cells. We're using them in the water industry to bring remote un-manned sites into the SCADA system that otherwise would be too expensive to hook up (no mains power on site, maybe no mobile GSM coverage etc).

    Just yesterday we were getting signals more than 10 miles, and also pushing signals through dense forests and woodlands.

    It works very well.

    Ultimately, the uptake of technologies such as LoRaWAN and Sigfox are going to be governed by the amount of data you're allowed to put into the air. We have concluded that LoRaWAN is not suitable for telemetry, because the restrictions on how much data you can transmit in 24 hours is (in our opinion) punitive, and not suitable for telemetry where, say for 23 hours of the day you are just sending routine event-based telemetry (pump on, pump off) but then the pump trips in hour 24 and you don't get notified because too many data events occurred during the day, and you are off the air because you have used your quota (LoRaWAN does not forward data into your network after the end-node has used up its allowance).

    To me, this smacks of the shenanigens played out by the mobile operators - making you pay for your data. It actually isn't, it's a means of maintaining space on the airwaves for all devices to have a chance to transmit their data, but as I mentioned, the allowance is not enough for serious commercial applications at the moment. Consequently we are running LoRa systems, but coming into our own, custom developed gateway.

    I hoping Weightless will gain more traction because it looks good, but at the moment, LoRa has the momentum (by a very long way) over all the others, though LoRaWAN will not be suitable (I believe) for most applications, unless you have VERY low amounts of data to send per day). I run three LoRaWAN gateways (on TheThingsNetwork), one in Durham and two in Aberdeen where data is forwarded from any device that wants to transmit on LoRaWAN. The gateways are based on Raspberry Pi's and work flawlessly.

    1. ZSn

      Re: Interesting article

      Does this also apply to the things network?

      1. ForthIsNotDead

        Re: Interesting article

        Yes. The three gateways that I'm running are all on TheThingsNetwork, so if your nodes are registered on the TheThingsNetwork, and my gateway(s) pick up your transmissions, the data from your nodes will be forwarded to you wherever you are in the world.

    2. A Dark Germ

      LoRa modems for me, dump LoRaWAN it's not secure.

      LoRa modems are my pick for FREE to PLAY with IoT

      I will not PAY to PLAY with IoT.

      Using LoRa modems not LoRaWAN.

      Can clone on LoRaWAN on TTN a year back.

      Poor SECURITY model, uses symmetric keys!

      Will not use LoRaWAN for bespoke projects will only use it for class demo to show how not to build SECURE systems..

      Adding 508a/608a from Atmel for hardware public key cryptography on our projects.

      SAML11 for the Secure boot IoT hub talking to your back-end via MQTT over TCP/IP.

      N.B. Even LoRa modems you need to apply the 1% rule here in UK/EU.

      Talk for 1 second wait for 99 seconds before use again.

      This is for fair sharing.

      I am in Forres, want to work on the IoT project with me?

      Covering 100 mile road Inverness to Aberdeen concept. for all Scotland IoT projects.

  3. doublelayer Silver badge

    I have doubts

    I looked into some of these a while ago--I had a system which could benefit from communications at a distance, where there was no WiFi. However, I couldn't justify it, as all the components I could find required a lot of investment in hardware at each point. When you keep in mind that the computing part of a lot of these were raspberry pis, it perhaps makes more sense why I wasn't exactly impressed by the $20 US LoRa boards I found. Of course, I'm not mass-manufacturing these things, but a lot of the IoT things used in industry could probably be set up in a way such that they don't need these. For example, factories and hospitals, as listed in the article, are environments where devices could easily use WiFi connectivity. They are indoors, so in a place that likely already has networking, and a place that has electricity sufficient for the machines to transmit on the comparatively power hungry WiFi. There are lots of cases where something like this could be necessary, but I seriously doubt it's going to replace devices that use more common and less expensive technology.

    Another issue is that open standards are a lot easier to develop for. If I build something that can use WiFi, I know it will work with the network infrastructure already in place. If I'm doing things locally, as usually these things will do, there is no data limit problem. Given that anything considering using something with maximum message lengths less than a kilobyte won't be sending video or something like that, WiFi connected to the wider network still won't have a data limit problem. Meanwhile, all these systems seem to require some type of central infrastructure, which is either provided at a lot of restrictions or has to be constructed by the user of the devices. It's a lot of work to do for most applications, so unless the range or power is absolutely necessary, I doubt they get used as often as the predictions estimate.

    1. Oneman2Many

      Re: I have doubts

      Did you read the article ? Its not designed for high data throughput, never claims to be. Its designed for low data volume sensors like temperature, water flow, light, distance, vibration that are sending a reading back once a minute. If you want to be sending high res video stream then these devices are not for you. However if you can process your video or images at the edge and send back the results of what you are looking for then there is potentially a use case for you.

      Its interesting you mentioned hospitals and factories. Both of which are terrible for WiFi signal and hard to cable for power and network. To put a traditional AP on a plant floor could easily cost > $10k. Compare that to a device that costs < $200, has a 10-year battery, can be set in resin to be completely environment proof, has a magnet on the back for rapid deployment. You could flood an entire plant floor with zero maintenance mesh network for less than the cost of a single AP.

      Going back to sensors, we are looking at a still image use case. There isn't enough bandwidth to send back the whole image but actually we can process the image at the sensor and send back a yes / no for what we are looking for. Problem we were hitting was the power used to do image processing was dropping battery life from 10 years to a few months. Solution was a rechargeable battery charging using a PV panel that worked from a nearby florescent light. Also as the solution is two way we can update the image processing code and send firmware updates though these can take days to get there.

      The issue over standards is still a pain but hopefully that should work itself out in the next couple of years.

      1. doublelayer Silver badge

        Re: I have doubts

        In fact, I did read the article. If you could reread my comment, please, you will perhaps notice that I mention video as a reason why data limits on WiFi wouldn't apply to devices such as this, as they will not be sending a lot of data.

        Hospitals and factories are very often highly networked. They are indoors, and there is electricity running everywhere. Machines, lighting, communications infrastructure, all are there. Especially with hospitals, there are also computers everywhere requiring network activity, not to mention patients who may have phones and laptops with them. Installing WiFi access points in an environment like this would make a lot of sense. FYI, a single WiFi access point doesn't cost anything like $200. A network covering a factory wouldn't be cheap, true, but depending on use case, it could be worthwhile. However, in a highly electrified location, many of the sensors could be directly wired.

        The use cases that make the most sense to me are those that need a long range where electricity and networking aren't available, most of which are outdoors. I can think of some use cases indoors as well. If you had a factory without WiFi and you only had one or two sensors, then installing WiFi only for those might have little point. However, if that were the case, I'd assume you could do that more efficiently by having those sensors attached to whatever they're sensing, as there are not that many of them.

        1. Oneman2Many

          Re: I have doubts

          As a comparison there are very few situations where you need video feed and even if you do then the chances are you will try and process it at the edge. The vast majority of IoT devices are going to be low data rates.

          As for factories, not cheap in the slightest to get infrastructure in there. From personal experience support industrial environment. > $10k for an access point is pretty common once you consider tapping into a 3 phase power supply, running fiber and never mind trying to do all this during maintenance windows. Same with hospitals, you generally don't want AP where there is infrastructure already. That verses putting in a couple of $200 battery power AP.

          Will add that WiFi is generally very difficult to do in a factory, lots of electric interference and metal getting in the way. Forget trying to get even a 3G signal in there. While is LaRaWAN is designed for longer range, there are plenty of use cases for it complementing WiFi.

          1. doublelayer Silver badge

            Re: I have doubts

            I will try one more time. I do not think, nor have I ever thought, that the type of devices we are discussing are the types to send video streams or other large data streams.

            Now to cover IoT vs IT: IoT is IT. It is a system that runs code, interacts with other devices, and in order to actually come under IoT, must connect to the internet at some point. It may not be exactly similar to other IT devices, and there may be other people to work on it, but it is technology that delivers information, and is more similar to IT infrastructure than infrastructure by most other departments. In addition, most of the things that are talked about as members of the IoT group as a whole are much more complex than something that runs some sensors and can communicate all the relevant information in some 255-byte chunks sent every once in a while. Even if we're limiting ourselves to equipment used in industrial things, there are things that require direct connection because they report information or receive commands in real time, and those things are always called IoT. In fact, if we wanted to split the broad category IoT, the devices better suited to the name would be those connected directly to the internet rather than sensors that use a gateway (of course, if that gateway also allows internet connections, then those sensors also belong to the IoT group).

            My doubts are still alive, but not because I fail to see use cases for the technology. I merely see a great deal of complexity in the technology that would make it difficult for some players to use it. Combined with the fact that WiFi exists and can be used by a lot of sensors, albeit with less effectiveness and many downsides (I admit these are there), many might not bother to implement it. This could cause the technology to stagnate.

            On the topic of hospitals, many of these already have WiFi. Part of the reasoning for this are patients who must remain there for some time, but the system is also used for wireless devices used by the staff. Those hospitals that lack it often still have wired internet, as they are some of the most computerized locations. I'll admit that I have less experience with factories, but I'd imagine that those for whom the term internet of things is a selling point are probably not strangers to internet-connected equipment on the factory floor.

      2. A Dark Germ

        They says years but its really hours.


        Total agree most people are fools, thinking IoT is IT.

        This is the problem poor education for people.

        Even the sellers or LoRaWAN in Scotland advertise WiFi on telephone poles with LoRaWAN @ Inverness uni talks. I pointed out it was a false image they did not know!

        I was asked to be quite about this point.

        I ran tests for Semtech way back 2016/17 on their own hardware with GPS tracking IoT LoRaWAN kit. It lasted a few hours if lucky.

        See my education site

        You see the GPS is a beast with power.

        When you think of a year long, this is a IoT device in deep sleep for 99.999% of the time.

        Then and only then can you have the coin cell last.

        So ten yeas might happen if the IoT is never used!

    2. A Dark Germ

      IoT is about sensors & actuators connecting to a IoT hub.

      Your missing the point of IoT.

      IoT is not IT in the field.

      IoT is all about sensors & actuators that send small amounts of data at periodic times or on request.

      Your confused . com like most people these days.

      Time to return to school i think.

      Please understand the problem domain before applying a solution.

      LoRa modems are free to use and can cover a vast range with small amounts of data less then 255 bytes.

      WiFi is IT, OK.

      Your off topic here 100%

  4. Anonymous Coward
    Anonymous Coward

    horses for courses

    I like this article as it agrees with my thoughts exactly.

    I've done a fair amount of experimentation with Lorawan and it's becoming part of my job now so I know a bit about it.

    I hear a lot of talk about which one will win out and people being wary of what LPWAN technology to back. But when it comes to NB-IoT vs Lora or Sigfox there really is no competition. They are for different things. NB-IoT will be great for time critical data and data where you need some kind of service guarantee. But it will be more expensive, there's no getting round the fact that mobile carriers will be paying for the air space and will have to make it bring in profit or else they wouldn't do it. It'll be much more suitable for small low powered devices than current mobile carrier offerings but it's always going to have a cost associated with it. The beauty of things like Lora or Sigfox is you can make your own LPWAN for a relatively low cost (and that cost just keeps getting lower). Lots of applications for IoT just don't have a requirement for strict real-time data or guaranteed delivery.

    I don't think the betamax/vhs analogy is appropriate, there's no need for one winner there will be multiple winners each with their own use-cases.

    1. Oneman2Many

      Re: horses for courses

      While there may be multiple solutions dependant on the use cases, most larger organisations don't want to be developing to different standards. The key would be presenting the data in a common way, not method for getting the data to the enterprise network and apps.

  5. Barleyman

    Available versus predicted

    Many / Most people commenting on IoT seem to be hobbyist of some sort or perhaps working on a single fixed installation. If you're designing something that needs to go out there and work in not-predefined locations reliably, LoRa is right out. SigFox actually has decent coverage depending on region but the data amounts are simply far too small for anything doing regular monitoring.

    LTE CAT-NB1 and LTE CAT-M1 are not here (in UK), period. No amount of handwaving gets around it, you design a device with either and you just made an expensive paperweight which might start working sometime in the future, or not. CAT M1 was "supposed" to work on existing basestations with just firmware updates but this didn't happen, surprise. In reality, you need hardware changes which means there's nonzero cost involved and operators don't wanna since there will not be that much business for some time. And since the network isn't there, business won't come.

    So for UK, for something with real coverage and reliable service, you've got two choices really. There's LTE CAT-1 which is a slightly cheaper-to-implement low speed (around 10Mbit/s) version of common garden 4G technology. It works with zero modifications on 4G networks, which is nice. What's not so nice is that you've now got a bona-fide cellular module that will do around £20-30 worth of damage when bought in volumes. Ouch. If you're deploying in US, you may very well get a reasonable M1 coverage which slashes module prices to one half or less.

    The fallback is the good old GSM GPRS. It's everywhere and it works. It's slow but the ~8kbit/s speed is a far cry above what you can achieve on SigFox and you can push data all day if you want unlike LoRaWAN. If you can get EDGE, you're in with roaring data rates.

    3G is on it's way out in intermediate term, better not bet on it for something that needs to be around for 5-10 years.

    So the sad truth is that if you need a serious IoT device deployed right now in UK, you've got to reach for a 20 year old cellular technology. Where's my flying car?

  6. Oneman2Many

    You need to look at your use cases, do you really need to be pushing MB of data or can you process that at the edge ? And cost of modem is peanuts compared to carrier costs.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like