IoT Link: LoRa vs. UNB. Part 2: Business
The second of a series of articles devoted to the description of the main differences between low-power long-range radio communication technologies that are now spread in IoT systems: LoRa broadband from narrowband (UNB, Ultra Narrow Band) systems, such as Sigfox and Strig, as well as their practical application .
Hi, Giktayms!
I apologize for the long silence - I promised many to post this part of the article on August 22nd, but I was dragged out a little by the beautiful city of Almaty , and then other urgent matters. As compensation, I promise, after today's text, to tell you about specific equipment for LoRa / LoRaWAN networks (ours and not only), as well as about some technical aspects of the operation of LPWAN networks. The last text would logically have to continue the very first one - there will be both about noise immunity, and about the problem of feedback, and about achievable speeds - but on the one hand, in our lectures many were interested in these details, and on the other, today's text was also promised more many, and delay it even longer quite ugly. Therefore, first talk a little about business.
')
So let's go. In the first part, I told you what the competing protocols for long-range IoT networks, LoRa, Swift, Sigfox, and others, are similar to and how physically different are the competing protocols.
But the differences at the physical level are academic knowledge, but we care about whether we can implement a particular project on a particular protocol. The client does not need to know the principle of the hammer - he needs to hang a picture on a nail.
And here everything becomes quite fun.
The main thing you need to understand about all the protocols listed is that they are all proprietary, that is, at a minimum, the physical level belongs to a specific company. Now an open standard Weightless is being developed (it belongs to the UNB-systems group), but before it’s practical implementation it’s not yet known how long it will be - and, in fact, this implementation itself is questionable: there are a lot of standards being developed, only a few survive.
The second important point: patent owners usually want to control not only the physical layer.
Actually, what levels do we have at all - if we consider not the OSI model, but the implemented IoT project?
The aggregate of all this forms the network for collecting the generation of the Internet of Things. And if we talk about the degree of openness for third parties of the three main technologies on the CIS countries market - LoRa, Strizh, Sigfox - then it looks like this:
Sigfox - historically the first large company in this market - chose a rather closed model for itself. Sigfox needs to buy base stations and conclude an agreement with it to deploy a network, to which paid access to third parties will be provided. Part of the money received from subscribers, under this contract will go to Sigfox. Base stations transmit data to Sigfox servers, that is, the communication channel is not physical, of course, but logical, and the upper level software also belongs to Sigfox.
Fortunately, Sigfox did not usurp the market of chips and end devices - he agreed with Texas Instruments, SiLabs and other manufacturers to support their network, so if you want to make or order some non-standard Sigfox device, this will not be a fundamental problem.
Unfortunately, the problem may be its network connection: if in Europe Sigfox managed to turn around, now, with the advent of LoRa, its expansion has actually stopped.
“ Strizh telematics ”, explaining the need for this underdeveloped market, went even further - in their case, the vertical of power has been brought to an absolute limit: the company itself produces base stations and end devices, provides the network server itself, and even prefers to install base stations . Each connected device is charged a subscription fee. The company does not manufacture the chips by itself - it uses ready-made transceivers Semtech and Axsem - however, it releases firmware for them, without which these chips in the network "Swift" are obviously completely useless.
Such a degree of closeness not only prevents many projects from being implemented at all, but even where implementation is possible, represents a serious risk for the business: you are tied to a single supplier for all levels of the solution . Step to the left, step to the right - escape, jump on the spot - attempt to fly away.
Perhaps, it is possible to agree with Strizh-Telematik for some kind of relaxation in their policies, but the company's website does not directly report this to anything.
Finally, LoRa . LoRa technology is owned by Semtech, a major semiconductor manufacturer. Actually, LoRa Semtech decided to do the same - to make money on semiconductor products, giving everything else to the mercy of consumers.
LoRa chips for end devices are available for free, documentation on them is open, everyone can make devices on them. Chips for base stations are sold only under a direct contract with Semtech, and the documentation for them is sold under NDA, but more than that, base station manufacturers are not limited, so this environment is also competitive, although due to the technical complexity of the BS, there are fewer players than among the end devices.
Moreover, a simple LoRa base station can be made on chips for end devices - it will have technical limitations, but they are not critical for many projects. However, about this - in the next article.
If you want to build a cellular network with multiple BSs - and this is the main business model for all LPWAN networks - then Semtech proposes using the standard LoRaWAN MAC layer. Source codes are available on github , the license is modified BSD. How you build a network, what you use as a network server, how you charge subscribers and if you charge them, if you provide them with some kind of cloud services or if you suggest using your own services - your problem alone. Semtech doesn't interfere in how you solve it. If you do not have enough qualifications to solve it, you can contact third-party end-device developers (for example, us), operators of LoRaWAN networks (we can introduce) and top-level software developers (many in b2b, for example, use Azure).
If you need an enterprise-wide network, you can do without LoRaWAN by building a star network with a coverage radius of two to three kilometers, fixed device speeds and an inexpensive base station, which does not require a network server or internet connection to work . For example, in the photo at the beginning of the article is the single-channel BS LoRa of our development, which stands in the basic configuration of only 6500 rubles.
It is this freedom of choice that determined the great interest in LoRa networks: different consumers have different needs and different business models, and it’s simply impossible to implement many of them under the severe restrictions of owners of Sigfox or Stryzh technologies.
If you separate projects by type in terms of network topology - there are three main options:
LPWAN Cellular Networks - City Scale Networks
These are networks with many base stations (for example, about 200 pieces are needed to cover Moscow), as a rule, served by the network operator, which provides little access to the owners of the end devices for money.
From a technical point of view, LoRa, Strizh and Sigfox in such networks are almost equal - for all three, the basic business model is aimed specifically at cellular networks. But from a business point of view, there is a significant nuance: Semtech (LoRa) does not regulate the work and tariff network of network operators, which theoretically allows them to compete with each other, and also to offer those conditions that they consider reasonable for subscribers. Sigfox establishes a rigid tariff grid; Strizh does not officially publish such a grid, but it is obvious that all negotiations on tariffs will have to be conducted with it directly.
However, the word “almost” was not used above by chance. LoRa and UNB-technologies are equal in terms of data delivery from the subscriber to the BS. But, as mentioned in the previous article, with the symmetry of the channel for UNB-technology, everything is pretty bad - data transmission from the BS to the subscriber is difficult, and sometimes simply impossible.
As a result, UNB systems are used only for data removal, for example, from metering devices, but another big topic is ASUNO, that is, street lighting control is inaccessible to them, because It implies an effective and continuous communication in the direction of the BS → subscriber.
Local Star Topology Networks
If your scale is a few kilometers maximum (for example, an enterprise, a business center, a holiday village, career development, etc.), then you most likely do not need a cellular network with all its difficulties in implementing and deploying. In this case, it will be cheaper and more convenient to use the star topology with one base station serving a hundred or two or three subscriber devices.
Sigfox does not allow such a model in principle. Not included in licensing options. Point, paragraph.
"Swift" theoretically allows, but with very significant business constraints. Since the Stryzh technologies are sharpened for cellular networks, even if you install one base station, the data from it will still go to the network server and Stryzh Telematics cloud services. There is information - on the company's website is not provided that the network server can be licensed for local installation. As a result, although the Stryzh business model in this matter suddenly turns out to be softer than Sigfox, in practice it is not as flexible and convenient as many clients would like.
LoRa, as mentioned above, in no way limits either the business model or even the use of LoRaWAN. Therefore, the local network on LoRa to a small territory and a small number of devices can be built as required by the customer - including completely transferring this network to him without any licenses and deductions. The base station itself can act as a server for such a network — for example, the base station in the photo at the beginning of the article has a built-in microcomputer with Linux and can broadcast LoRa traffic to the MQTT protocol that many people understand.
Specific Topology Networks
These include, for example, radio relay networks.
Imagine power lines. On the one hand, power engineers want to control several basic parameters with an acceptable spatial resolution — wire tension (increases during icing), voltage on the wire (decreases when cut off by aborigines), corona discharges (occur on insulators when they are polluted). Put the sensors - not a problem, but what to do next with the data from them? 3G? Expensive, and the coverage of cellular networks is, to put it mildly, far from everywhere where power lines pass. Plc? With connection to a power line, distances of tens to hundreds of kilometers and a lot of transmitters? No one will seriously discuss this.
Meanwhile, if we have available an inexpensive transceiver operating in a license-free range with a range of kilometers, then technically nothing should disturb us by placing such transmitters on power lines equipped with sensors and transmitting telemetry between them along the chain. Having received a cheap unlicensed radio relay.
As you might guess, such networks can only be done by LoRa - where we can stop using LoRaWAN and everything associated with it, and write our own data transfer protocol on top of the LoRa physical layer.
Neither in Sigfox, nor in “Strizhe”, nor in the world of narrowband UNB systems at all, such a network topology is technically impossible.
Unwired Devices develops and manufactures communication modules for 6LoWPAN cellular networks and LoRa long-distance networks, as well as sensors and other terminals for these networks, including both hardware and firmware, supporting the necessary network technologies. In the case of LoRa networks, we develop all possible topologies: cellular and static radio relay networks, star-based object networks from a single BS, and devices for LoRaWAN global networks.
- IoT Link: LoRa vs. UNB. Part 1: Physics
- IoT Link: LoRa vs. UNB. Part 2: Business
- IoT Link: LoRa vs. UNB. Part 3: technical details
- IoT Link: LoRa vs. UNB. Part 4: LoRa networks and equipment
Hi, Giktayms!
I apologize for the long silence - I promised many to post this part of the article on August 22nd, but I was dragged out a little by the beautiful city of Almaty , and then other urgent matters. As compensation, I promise, after today's text, to tell you about specific equipment for LoRa / LoRaWAN networks (ours and not only), as well as about some technical aspects of the operation of LPWAN networks. The last text would logically have to continue the very first one - there will be both about noise immunity, and about the problem of feedback, and about achievable speeds - but on the one hand, in our lectures many were interested in these details, and on the other, today's text was also promised more many, and delay it even longer quite ugly. Therefore, first talk a little about business.
')
So let's go. In the first part, I told you what the competing protocols for long-range IoT networks, LoRa, Swift, Sigfox, and others, are similar to and how physically different are the competing protocols.
But the differences at the physical level are academic knowledge, but we care about whether we can implement a particular project on a particular protocol. The client does not need to know the principle of the hammer - he needs to hang a picture on a nail.
And here everything becomes quite fun.
The main thing you need to understand about all the protocols listed is that they are all proprietary, that is, at a minimum, the physical level belongs to a specific company. Now an open standard Weightless is being developed (it belongs to the UNB-systems group), but before it’s practical implementation it’s not yet known how long it will be - and, in fact, this implementation itself is questionable: there are a lot of standards being developed, only a few survive.
The second important point: patent owners usually want to control not only the physical layer.
Actually, what levels do we have at all - if we consider not the OSI model, but the implemented IoT project?
- Physical level In our case, it is a semiconductor chip that implements the communication channel we need, and / or a special firmware for a certain set of typical chips, if the communication channel can be implemented purely in software.
- Communication protocol This is the MAC layer and the network layer of the OSI model — end devices transmit data to the base station of the network using them.
- Actually end devices. There is a chip in them that implements the necessary communication channel with the necessary MAC and network layers.
- Base station.
- Data network. These are both physical communication channels from the base station to the central server of the network, and the protocol running over them.
- Control system. A central network server that receives data from base stations. It may either include the top-level software that is engaged in the storage-processing-visualization of this data, or simply transfer them somewhere further.
The aggregate of all this forms the network for collecting the generation of the Internet of Things. And if we talk about the degree of openness for third parties of the three main technologies on the CIS countries market - LoRa, Strizh, Sigfox - then it looks like this:
Sigfox - historically the first large company in this market - chose a rather closed model for itself. Sigfox needs to buy base stations and conclude an agreement with it to deploy a network, to which paid access to third parties will be provided. Part of the money received from subscribers, under this contract will go to Sigfox. Base stations transmit data to Sigfox servers, that is, the communication channel is not physical, of course, but logical, and the upper level software also belongs to Sigfox.
Fortunately, Sigfox did not usurp the market of chips and end devices - he agreed with Texas Instruments, SiLabs and other manufacturers to support their network, so if you want to make or order some non-standard Sigfox device, this will not be a fundamental problem.
Unfortunately, the problem may be its network connection: if in Europe Sigfox managed to turn around, now, with the advent of LoRa, its expansion has actually stopped.
“ Strizh telematics ”, explaining the need for this underdeveloped market, went even further - in their case, the vertical of power has been brought to an absolute limit: the company itself produces base stations and end devices, provides the network server itself, and even prefers to install base stations . Each connected device is charged a subscription fee. The company does not manufacture the chips by itself - it uses ready-made transceivers Semtech and Axsem - however, it releases firmware for them, without which these chips in the network "Swift" are obviously completely useless.
Such a degree of closeness not only prevents many projects from being implemented at all, but even where implementation is possible, represents a serious risk for the business: you are tied to a single supplier for all levels of the solution . Step to the left, step to the right - escape, jump on the spot - attempt to fly away.
Perhaps, it is possible to agree with Strizh-Telematik for some kind of relaxation in their policies, but the company's website does not directly report this to anything.
Finally, LoRa . LoRa technology is owned by Semtech, a major semiconductor manufacturer. Actually, LoRa Semtech decided to do the same - to make money on semiconductor products, giving everything else to the mercy of consumers.
LoRa chips for end devices are available for free, documentation on them is open, everyone can make devices on them. Chips for base stations are sold only under a direct contract with Semtech, and the documentation for them is sold under NDA, but more than that, base station manufacturers are not limited, so this environment is also competitive, although due to the technical complexity of the BS, there are fewer players than among the end devices.
Moreover, a simple LoRa base station can be made on chips for end devices - it will have technical limitations, but they are not critical for many projects. However, about this - in the next article.
If you want to build a cellular network with multiple BSs - and this is the main business model for all LPWAN networks - then Semtech proposes using the standard LoRaWAN MAC layer. Source codes are available on github , the license is modified BSD. How you build a network, what you use as a network server, how you charge subscribers and if you charge them, if you provide them with some kind of cloud services or if you suggest using your own services - your problem alone. Semtech doesn't interfere in how you solve it. If you do not have enough qualifications to solve it, you can contact third-party end-device developers (for example, us), operators of LoRaWAN networks (we can introduce) and top-level software developers (many in b2b, for example, use Azure).
If you need an enterprise-wide network, you can do without LoRaWAN by building a star network with a coverage radius of two to three kilometers, fixed device speeds and an inexpensive base station, which does not require a network server or internet connection to work . For example, in the photo at the beginning of the article is the single-channel BS LoRa of our development, which stands in the basic configuration of only 6500 rubles.
It is this freedom of choice that determined the great interest in LoRa networks: different consumers have different needs and different business models, and it’s simply impossible to implement many of them under the severe restrictions of owners of Sigfox or Stryzh technologies.
If you separate projects by type in terms of network topology - there are three main options:
LPWAN Cellular Networks - City Scale Networks
These are networks with many base stations (for example, about 200 pieces are needed to cover Moscow), as a rule, served by the network operator, which provides little access to the owners of the end devices for money.
From a technical point of view, LoRa, Strizh and Sigfox in such networks are almost equal - for all three, the basic business model is aimed specifically at cellular networks. But from a business point of view, there is a significant nuance: Semtech (LoRa) does not regulate the work and tariff network of network operators, which theoretically allows them to compete with each other, and also to offer those conditions that they consider reasonable for subscribers. Sigfox establishes a rigid tariff grid; Strizh does not officially publish such a grid, but it is obvious that all negotiations on tariffs will have to be conducted with it directly.
However, the word “almost” was not used above by chance. LoRa and UNB-technologies are equal in terms of data delivery from the subscriber to the BS. But, as mentioned in the previous article, with the symmetry of the channel for UNB-technology, everything is pretty bad - data transmission from the BS to the subscriber is difficult, and sometimes simply impossible.
As a result, UNB systems are used only for data removal, for example, from metering devices, but another big topic is ASUNO, that is, street lighting control is inaccessible to them, because It implies an effective and continuous communication in the direction of the BS → subscriber.
Local Star Topology Networks
If your scale is a few kilometers maximum (for example, an enterprise, a business center, a holiday village, career development, etc.), then you most likely do not need a cellular network with all its difficulties in implementing and deploying. In this case, it will be cheaper and more convenient to use the star topology with one base station serving a hundred or two or three subscriber devices.
Sigfox does not allow such a model in principle. Not included in licensing options. Point, paragraph.
"Swift" theoretically allows, but with very significant business constraints. Since the Stryzh technologies are sharpened for cellular networks, even if you install one base station, the data from it will still go to the network server and Stryzh Telematics cloud services. There is information - on the company's website is not provided that the network server can be licensed for local installation. As a result, although the Stryzh business model in this matter suddenly turns out to be softer than Sigfox, in practice it is not as flexible and convenient as many clients would like.
LoRa, as mentioned above, in no way limits either the business model or even the use of LoRaWAN. Therefore, the local network on LoRa to a small territory and a small number of devices can be built as required by the customer - including completely transferring this network to him without any licenses and deductions. The base station itself can act as a server for such a network — for example, the base station in the photo at the beginning of the article has a built-in microcomputer with Linux and can broadcast LoRa traffic to the MQTT protocol that many people understand.
Specific Topology Networks
These include, for example, radio relay networks.
Imagine power lines. On the one hand, power engineers want to control several basic parameters with an acceptable spatial resolution — wire tension (increases during icing), voltage on the wire (decreases when cut off by aborigines), corona discharges (occur on insulators when they are polluted). Put the sensors - not a problem, but what to do next with the data from them? 3G? Expensive, and the coverage of cellular networks is, to put it mildly, far from everywhere where power lines pass. Plc? With connection to a power line, distances of tens to hundreds of kilometers and a lot of transmitters? No one will seriously discuss this.
Meanwhile, if we have available an inexpensive transceiver operating in a license-free range with a range of kilometers, then technically nothing should disturb us by placing such transmitters on power lines equipped with sensors and transmitting telemetry between them along the chain. Having received a cheap unlicensed radio relay.
As you might guess, such networks can only be done by LoRa - where we can stop using LoRaWAN and everything associated with it, and write our own data transfer protocol on top of the LoRa physical layer.
Neither in Sigfox, nor in “Strizhe”, nor in the world of narrowband UNB systems at all, such a network topology is technically impossible.
Unwired Devices develops and manufactures communication modules for 6LoWPAN cellular networks and LoRa long-distance networks, as well as sensors and other terminals for these networks, including both hardware and firmware, supporting the necessary network technologies. In the case of LoRa networks, we develop all possible topologies: cellular and static radio relay networks, star-based object networks from a single BS, and devices for LoRaWAN global networks.
Source: https://habr.com/ru/post/397319/
All Articles