Big Data for M2M – too much too soon

October 27, 2013
Walker Information

Walker Information

The M2M aka Internet of Everything market is developing quickly. It is interesting to see how analysts, vendors and journalists over the last few months collectively shifted from one big number to another – from number of connected devices to Big Data business potential for Internet of Everything. It is a sign of growing up that we leave the connection focus and huge abstract estimates on number of connected devices for a focus on data and information. Data is the gold of M2M and I’m pleased that the industry focus is shifting towards the information. The industry loves buzzwords and in the M2M case we ended up in the Big Data bucket. ABI just estimated that Big Data and Analytics in M2M will generate revenues of $14 Billion in 2018. We are in the beginning of M2M aka Internet of Everything and organizations are starting to gather useful information from sensors and things. We see more open API:s and people have started blending the collected M2M data with other data to enrich the value of the information. But few if any are even close to a situation which Big Data is addressing. Most data collected in M2M applications are a couple of bytes from time to time. When mobile networks are used its predominantly tiny data in 2G networks, a lot of narrow band wireless sensors are used and when POTS is used it is typically a couple of seconds transmission. And it is not likely to change! Most of the data we are collecting from things are meter reading data, positions, status information, health data, times and so on. Nothing of that comes close to Big Data. But with big volumes of sensors and transactions we will rather need sophisticated decision support systems.

So why are Big Data brought into the M2M aka Internet of Everything discussions? First and foremost people understand that focusing on number of connected devices isn’t interesting any longer and since the solutions and applications most often are industry, function or company specific it is really difficult to translate them into mind-blowing numbers. But by looking at it from a Big Data point of view new huge market numbers are the results. Secondly, as we become more data oriented the IT companies are getting involved in Internet of Everything. And since many of them are preaching the need for Big Data solutions it is easy to bring the “M2M stuff” into the Big Data story.

I see a risk that the Big Data twist on M2M will make organization miss the ball in their own M2M efforts. Volumes, velocity and variety is not among the key issues in M2M projects today. It is too much too soon. In my experience the best approach is to get going, connecting a couple of things together with a relevant specialist M2M Service Enabler who has most of what is needed already. That way we will learn rapidly from real-time data in our own business and the trials are rapid and affordable. Then its quite easy to see what data to collect from which sources and step by step develop the collecting, management and distribution of data for maximum value to the organization. It is only when the information appears in business systems, business processes, decision-making systems or user applications that the value is realized. We simply don’t need any Big Data methods or solutions for this. It is obvious though, that sooner or later, especially if streaming data is involved, organizations will have a lot of data to process. And M2M applications will of course increasingly add to these systems down the road.


Alternative WAN solutions for Internet of Things

December 7, 2012

SIGFOX LogoAt the local area network (LAN) level we have a lot of different wired and wireless alternatives for M2M connectivity. But at the wide area network (WAN) level we have few options. There are a lot of legacy solutions using the fixed switched phone network (PSTN and fixed broadband Internet access is the dominating way to connect LANs to the Internet. The mobile industry is positioning themselves as the obvious solution for wireless WAN connectivity. The mobile operator alternative includes the cheaper and low capacity 2G option, 3G and the emerging high-capacity and low latency LTE option. 2G is still by far the most utilized option with more than 90% of all M2M subscriptions. Satellite communication provides an almost complete outdoor coverage and is a frequently used alternative especially for tracking.

But there are potentially other alternatives for M2M WAN connectivity. Entrepreneurs are working on the idea to build a dedicated M2M network designed to connect billions of devices in a cost efficient and high quality fashion. One of the most interesting today is Sigfox in France who has designed and built a wireless network optimized for M2M using ultra narrow-band modulation techniques. They started roll out earlier this year and plan to have France covered by the end of this year which is amazing. Sigfox uses unlicensed spectrum (868 MHz in Europe and 915 MHz in the US) normally used by cordless phones. With open sight distances up to 40 km covered and when compared with GSM, for the same level of coverage, Sigfox’s solution requires around 1,000 times less antennas and base stations. The impact on cost is massive – it is 100 times less expensive to build, install and operate. They claim they will have France covered with some 1000 transmission sites. The radio modules embedded in the connected things are tiny and consume 1/50 of the power typically consumed by a cellular M2M module. With such low power consumption batteries could last up to 20 years before recharging or replacement is needed.

The Sigfox network is designed to connect millions of devices that only send messages occasionally – maybe once a week or once a year. The position of the object is included and data is encrypted. The bandwidth is only 100 bps which allows transfer of only small messages. And this will likely be the most common type of connected object why volumes could be very large and economies of scale could help Sigfox bring down cost to a couple of dollars per module. They predict that their efficiencies in running the network will enable them to connect devices for a couple of dollars a year.

Now you might wonder what type of mushrooms they have down in Toulouse. But already now they have announced that Clear Channel Outdoor Holdings is a customer using Sigfox to connect their billboards and MAAF Assurances just announced an agreement for their innovative connected objects’ household protection service, they have suggested to ETSI to make their proprietary UNB technology a standard and in September Intel Capital led their €10 million B round.

Friends, this is for real and companies like Sigfox have the potential to change many games onwards. Connected objects without hassle for a couple of dollars a year sounds attractive, doesn’t it? Bonne chance!

What will be the impact on M2M of announcement to shut down 2G?

August 8, 2012

AT&T just announced that they will shut down their 2G network no later then January 1 2017. They want to free up spectrum for mobile Internet network capacity. “Well in advance of this change, we will reach out to our relatively small percentage of 2G customers and offer them options to meet their needs” says a spokesperson to Fierce MobileIT. According to Berg Insight AT&T had 13.1 million M2M subscribers by the end of 2011. It is often stated that the 2G share of the M2M subscriptions are bigger then 90%. I don’t know the AT&T mix but it is likely that at least half of their 13 million subscribers use 2G.

PricewaterhouseCoopers recently made a study and found that 60 percent of wireless operators plan on decommissioning their legacy networks over the next five years. Their study would suggest that more operators would like to make similar announcements as AT&T.

M2M terminals are often expected to be operational 5-10 years which makes the choice of network technology difficult. From a capacity point of view 2G is often enough and 2G modules are cheaper than modules supporting 3G and 4G/LTE. Coverage of the 3G networks are often not as good as the 2G networks. It is far too early for LTE networks to take over. Support for circuit switched connections which are required when moving terminals from the fixed phone networks (PSTN) to mobile networks is only available in 2G. It will be interesting to see what AT&T’s announcement will do to customers, regulators and other mobile operators. And what about countries were mobile networks are all 2G?

I believe this is a challenge for the development of the M2M market. As far as I understand it, AT&T will have to subsidize the remaining 2G M2M customers to replace terminals or loose the customers to other operators. This new situation for ongoing and planned M2M projects might delay or in worst case kill them if the up-front investment becomes substantially bigger. This news also explains why M2M customers need over-the-air provisioned SIMs.

ETSI M2M standards and the oneM2M initiative

May 3, 2012

An absolute requirement for M2M to really take off is a common architecture and a set of global standards. A robust common service layer is needed. The standardization organization ETSI started en effort to create such architecture and standard in January 2009 and has started to deliver. ETSI has released a multi-service layer enabling an easier and efficient access to the telecommunications network functionalities, designed to support a wide range of services over different underlying networks. It has to be noted that the work is IP only why issues like moving fax machines and other analog devices from PSTN to mobile networks isn’t covered.

According to Mr Enrico Scarrone, Chairman ETSI TC M2M, they try to build the standard on existing and available bits and pieces and focus on identifying gaps and fill them. At M2M Forum in Milan he presented also the oneM2M initiative which is a global partnership project aiming at removing the barriers for the Internet of Things created by local and industry related differences and approaches. We all know the power of 3GPP which strongly supported the developments of the third and fourth generations of cellular networks. Hopefully oneM2M will pave the way for a rapidly growing Internet of Things where data from different industries can be combined to create new innovative solutions and business models.

Local M2M Gateways

April 12, 2012
The mighty mobile industry is a major force behind M2M today. They have what it takes to connect things, they need to find growth beyond the six billion active SIM cards today and they are promoting the concept of M2M aggressively. But at the same time it is quite obvious that a lot of things will be connected without SIM card. We already have a lot of connected devices in the PSTN network, many electricity meters are connected using PLC or other wireless network technologies than the mobile networks and wireless technologies like W-Fi, Bluetooth, ZigBee and RFID are already used in many applications. Different technologies have their strengths and weaknesses, new ones are coming to market while old ones are fading away. But the concepts of wide area and local area networks remain. Generally speaking it takes technologies optimized for LAN or WAN to build cost and capacity efficient networks. Sometimes even a shorter range network structure is needed like a Personal Area Network.I believe the mobile networks will be used for some connected devices, especially moving things or if one want to avoid dealing with Firewalls. But more so as the preferred WAN solution for things connected to a local area network solution. I have come across estimates that one out of ten connected devices will have a SIM card and I think that could be a reasonable estimate. If Ericsson’s “50 Billion connected devices by 2020” would be reality, 5 Billion of them would have a SIM card. That is a lot more SIM cards than the 108 Million mobile M2M SIM cards that Berg Insight estimate are active today world-wide. But what about the other 45 Billion connected devices? How will they be connected and managed and how will relevant data generated by these devices become easily available for application developers and integrators?

The most immediate challenge for the M2M industry is to establish a rich assortment of M2M services enablers in order to make development and maintenance of M2M application more resource and time efficient. M2M Service Enablers will have different features and specialities and they can be deployed in three different ways: on top of operator connectivity services, as in-house solutions or in independent service providers.

But another very interesting area to be addressed is how to connect devices in a local infrastructure in order to enable resource efficient development, maintenance and monitoring as well as a structured way to deal with relevant data. I use to refer to “Local M2M Gateways” and I have started to look for clever ideas and solutions in the market. I am convinced there is room for a whole range of different products optimized for different situations, still providing a quite standardized interface to M2M Services Enablers. In some cases we need to connect locally using only one technology and in other situations we need to support a mix of several technologies. Connecting sensors or things with sophisticated embedded systems put different requirements on the Local M2M Gateways. The choice of WAN-connection, with or without backup, is yet another area where we will need different solutions. And whichever solution we end up using, it has to be cost efficient, easy to deploy and maintain and robust. These will be important tools when helping organizations to design relevant M2M solutions to meet their challenges and opportunities.

The 2G/3G dilemma

January 29, 2012

Today some 95% of the mobile M2M connections are 2G. It is absolutely natural since the functionality and capacity needed for most of today’s applications is fulfilled in 2G and the modules are substantially cheaper. An off the shelf 2G SIM card would cost around $3-4 per month plus $0,5-2/MB data transmitted (in Sweden) and CSD, GPRS and SMS is enough for most M2M applications today. But there are dark clouds in the horizon! We don’t know how far away or how fast they come forward, but they are definitely there. Let’s try look at what these clouds contain.

A typical M2M deployment would count on terminals to be in service more than five years, often 10 or even 15 years. That’s long time! It is about 20 years since the GSM services came to market, and betting on the same networks to still be there with great coverage and good service might be something to consider carefully. A customer service person at my previous mobile operator in Stockholm told me: “sorry but we don’t invest in the 2G network anymore”. One of my companies, Possio, help mobile operators to move analog devices from the fixed network, PSTN, to mobile networks using primarily circuit switched connections in 2G (CSD). They experience operators, one by one, deciding not to introduce any new CSD based services in their network. They keep the existing ones, but obviously not for ever. I believe CSD will not disappear over night but this is worth looking into when making the bets

The connect part of M2M is the least interesting and rewarding. It is the compute part that makes the difference. IP is today, by all means, the dominating communication platform across all industries. The IP development environment is solid and rich, application support endless and skill is really everywhere, from developers to support people. An M2M bet today should in most cases be built on IP and one should really try understand if performance in 2G GPRS/EDGE will be enough for making all wanted computing during the life cycle. It is easy to foul yourself when it comes to performance and capacity. My first business trip with IBM went to Copenhagen 1983 where serious old men unanimously stated that with this capacity nothing is stopping us any longer. This was an ISDN conference.

The end-of-life problem is always something to take into account. Module manufacturers normally bring to market new pin compatible modules for their most popular models. But one day they will issue an end-of-life notice and then it is last order date and finally the spot market to rely on before it is over. In other words, when a market decrease it’s a chicken race between the module vendors. They not only want to understand how fast the market disappears (remember they have good numbers to watch) but they also want to ensure the best moment to bring their customers forward on a new platform and not lose them to a competitor.

The cost of modules for 2G or 3G differs a lot. As of today a 3G module would be roughly double the price of a 2G module and the difference could be $25-30. That is a lot especially if you need many. But it is important to look at the entire cost envelope, both capex and opex, over time. The cost of the actual deployment is normally high since it takes human beings to prepare the installation, to ensure other people involved are available, to get and verify permission for entrance and finally to go on site. Each installation is obviously different depending on industry, security levels, distances, type of application, etc but it can easily take a couple of hours per terminal which would translate into hundreds of dollars. One of my companies is active in retail environments where they often experience a lot of problems especially with access permission and coordination with other people needed (electricians, operator staff, alarm staff, etc). This is why we need to get it right first time – we can’t afford to go back – and why the installations will have to be operational for many years. When planning an M2M solution this might well be the most important aspect of the business case and the biggest risk for failure.

I believe this question – should I stay or should I go – is very important for all of us in the M2M business. There are no generic answers to the question about going 2G or 3G but it seems inevitable that sooner or later 3G will be the primary network why focus and investments thus quality and coverage in 2G networks will erode. How fast this happens is of course also depending on geography. In order to put more light on this important question I will ask a couple of knowledgeable individuals from within the industry about their views and post them here.

Networks for M2M today

December 14, 2011

What networks are used for M2M applications today? We know a lot of applications use the fixed network (PSTN) but I lack data on that. I visited a friend with a quite small but modern grocery store the other day and beyond the main connection provided by headquarters for cash registers, computer systems, telephony etc, he has three separate PSTN lines to his shop: one for a video surveillance system, one for recycling machines and one for an alarm. When talking to a friend at a fixed line operator recently he mentioned that they have energy companies with water power plants with thousands of PSTN subscriptions for level gauges.

The 2G networks are by all means the most utilized mobile networks for M2M today. Beecham Research claims over 95% of M2M applications use 2G today. Good coverage, reasonable prices, limited capacity required, affordable modules and history explains this situation. And despite push for 3G from vendors and operators, not a lot is happening except for some specific applications. Vendors obviously want to sell new products and some operators want to re-use 2G spectrum or move users for other reasons. For people developing or operating an M2M solution a key question must be to understand when the time is right to change from 2G to 3G (if ever) and how to implement the change. A clear detailed roadmap could be a competitive edge for an operator. Interestingly enough some, primarily vendors and operators, already claim 4G is the way to go for M2M.

Most of Ericsson’s modules business ended up in ST-Ericsson some years ago and they re-entered in 2007 to provide cost effective broadband modules to be built into PCs. Quite silently they announced the end of their broadband modules business in December 1, 2011. The reason provided for the exit was: “our position on the market does not provide the scale we need to achieve the desired profitability”. I believe the actual size of the market for 3G modules also is part of the explanation. ABI claimed last summer: “USB configurations are outselling embedded modems by a ratio of more than three to one” and expected USB configurations to stay bigger until at least 2017.

3G has been around for some ten years now and in dense populated areas it is heavily utilized and 4G is starting to be deployed. But in rural areas most people relay on 2G (some even don’t have 2G) and in some developing countries they haven’t even started to deploy 3G yet. In the end of the day it is market requirements that decide when and to what extent networks will be built, upgraded and utilized, not operator or vendor desire. Still today most M2M applications don’t need much more than decent connectivity (I don’t include Pads, tablets and Smartphones in M2M), and for these to go 3G we need competitive coverage, prices and module prices compared to 2G.

M2M over fixed lines (PSTN)

November 22, 2011
Machine-to-Machine solutions have been around since analog modems came to market and millions of elevators, alarms, nurse phones, vending machines, franking machines, fax machines, recycling machines and level gauges utilize the fixed telephone networks (PSTN) to communicate. I have been trying for years to collect data on the number of machines connected to PSTN but it is information very hard to find. Operators typically don’t know what is behind the first socket of a PSTN installation and it is common that phones and machines share subscription.The two key limiting factors for M2M over PSTNare obviously that machines have to be connected by wire and the cost involved. The PSTN subscription alone is typically 100-300$/year which immediately prevents massive roll-out. The situation differs from country to country but it is often so that cost for new installations has to be carried by the subscriber and in some markets it takes forever to get a new subscription.Mobile networks rapidly grew to cover most of the geographies, modules were developed for none-phone usage and some operators started quite early on to build a new wireless generation of M2M solutions. This together with the massive Internet forces created a mobile M2M hype around year 2000 and our VC, BrainHeart Capital, invested in Wireless Maingate and Wireless Car at the time. I think most would agree that the ideas were great but it was far too early since technology, networks and services where not ready enough to fly.

GSM was developed to support packet switched data communications (GPRS) but a quite well hidden secret is that GSM also supports the circuit switched data communications (CSD) used in the traditional fixed phone networks (PSTN). CSD made it possible to move terminals from PSTN to GSM-networks which still is an attractive approach in networks where CSD is enabled. Typical usage has been encrypted phones and mobile fax. Still today this is an attractive way to move fixed terminals to a mobile network to save money, add flexibility, enable wireless offices or to enable fixed line operators to remove parts of PSTN that never can be upgraded to broadband and/or don’t have subscribers enough to carry the infrastructure cost. There is an important  difference between moving an existing PSTN terminal to a mobile network and to replace an existing PSTN terminal with an IP-enabled mobile terminal. Reasons for moving terminals include taking advantage of made investments in products and education, a desire to make changes step by step and last but not least that some applications like fax is technically very difficult if not impossible to do over a packet switched network (like GPRS). On top of technical arguments are things like the need to call a place and not an individual. Mobile phones are personal and fixed phones often shared.

The PSTN networks are on their way to be replaced by mobile networks but the situation differs a lot from country to country. In Sweden TeliaSonera has started to take down PSTN in rural parts of the country replacing the subscriptions with mobile alternatives. At the same time subscribers leave their fixed phone subscriptions behind and it was recently proposed that the concept of area codes are taken away to mirror the fact that the fixed phone is going away. The number of fixed line subscribers in Sweden are 2,5 million, almost half of ten years ago, compared to 13 million mobile subscribers. The concept of mobile one phone offices once invented by Spring Mobil in Sweden is now very popular in Northern Europe. By moving the switch to the network and removing the fixed phone infrastructure companies save a lot of cost and increase accessibility and flexibility. But the fixed infrastructure was often used also for other things that has to be taken care of including fax, door opening systems, conference phones, alarms and franking machines. I think the one phone offices will continue to spread and this will be the single biggest market for moving terminals from PSTN to mobile networks over the next couple of years. But it has to be underlined that it is not a simple thing to do since both fixed and mobile networks behave differently and many of the fixed network devices use odd protocols for communication. Most difficult of all is fax since it uses a very sensitive protocol.

In developing countries we often see fixed networks with limited coverage and mobile networks gaining momentum due to speed and cost of deployment. Most of these networks are still 2G and in combination with extensive use of paper mobile fax is a popular CSD application. But generally speaking I think GPRS will be the predominant connectivity method for M2M applications in developing countries many years ahead.

Moving devices from PSTN to mobile networks is an important part of the M2M industry. Since it seems like nobody has good enough data on the size of the M2M over PST market I would appreciate any data you could share with me on connected machines to PSTN and I hope one day to be able to share a decent overview helping us all to address this part of the M2M market which I often refer to as the narrow band data opportunity.

%d bloggers like this: