Microsoft and Google recently announced their HealthVault and GoogleHealth offerings as they vie for medical M2M ownership. What these offerings fail to acknowledge is the need for highly optimized, end-to-end platforms, including simple-to-set-up, cheap, secure data acquisition devices, if M2M applications such as healthcare are to grow exponentially.
For years the M2M market has largely been an area of idle curiosity, albeit a very lucrative one for a number of consulting companies. It's personified by interesting highly vertical applications like truck-fleet management or trackside monitoring of Formula One race cars.
Remote healthcare, traffic systems (road tolling), and energy monitoring - key application areas for M2M - are all about data. In essence, M2M involves retrieving data from an "asset", whether it's the health parameters of an individual, the where and how a vehicle is being driven, or the energy consumption of a washing machine. M2M networks are all about making a reliable connection to the asset to acquire the data, collecting the data on a central database, and then analyzing it. Equally important, though often forgotten, is the feedback based on the analysis, whether that is personal medical advice, a road-tolling bill, or controlling when the washing machine is allowed to run.
Because M2M networks are perceived as ultimately being about data, companies like Microsoft and Google see themselves as potential market leaders. However, the buzz around these product announcements obscures what has been one of the major stumbling blocks for the deployment of massive M2M networks - the physical mechanism of retrieving data from assets. Most of today's M2M deployments typically number in the hundreds or low thousands of users and are based on proprietary systems and hardware. However, the potential market for healthcare deployments, for example, is not in the tens of thousands, but in the millions, if not hundreds of millions. These deployments are not going to be achieved purely by offering a well designed database and middleware APIs, but rather by offering an end-to-end platform, including the data acquisition/communication pieces as well.
Missing the point
One misconception concerning M2M networks is the failure to understand that most of the assets that will require monitoring are not going to want to do it via a PC platform. Nor are they going to incorporate an embedded PC.
For one thing, it's too expensive. In healthcare, for example, the required data is likely to be as simple as a pulse-rate, temperature measurement, or the output from a bed-occupancy sensor - "occupied" or "not occupied". There's an important distinction to be made here. Exciting assets like monitoring and tuning the engine parameters of a Ferrari on a race course can afford expensive data collection and transmission solutions; every day assets can't.
Another factor is that assets will have different communication needs. Medical devices will want to communicate wherever they are, so they probably need to link via a mobile phone. In the case of assisted-living sensors, they'll need to talk to a cheap and simple home gateway. The same is true for domestic energy monitors. Automobiles will talk to roadside sensors using short range wireless networks.
The majority of these devices are unlikely to incorporate a cellular or conventional broadband link themselves. Instead they will rely on a lower-cost, lower-power wireless connection to a gateway device, which may be a residential box, municipal network, or mobile phone.
In short, if M2M applications such as healthcare are to explode, the assets need to have a way of communicating that is secure, simple and cheap.
"This is why the pronouncements of the likes of Google and Microsoft are missing the point," says Nick Hunn, CTO of Ezurio. "Their offering of a secure database, although an important part of the solution, is just one component part of it. To expand usage from the current low installed base, companies that want to become significant players need to understand that they must support a complete platform that extends from the asset's data all the way to the secure database, which then feed the enterprise or public application above it."
Moreover, adds Hunn, "That real platform needs to be designed to be simple enough to work with low-cost hardware, as many of the asset sensors will be based around inexpensive microcontrollers - a very different concept from that of the Linux or Windows powered world that some large corporations seem to believe is the only one that exists."
Hidden costs
However, there are a lot of unresolved real-world problems remaining in connecting the assets to the enterprise applications at low costs, whether via Wi-Fi-type connections or cellphone-type connections.
For example, the mantra that wireless M2M-network nodes will all be connected to the internet and will cost less than $2 each tells only part of the story. Yes, the semiconductor industry can provide a $2 radio - but that's a ZigBee-style 802.15.4 radio, not a radio that can use cellular networks. So while large subnets of ZigBee- or WiBree-based networks can tie large networks of asset nodes together, the system still needs to get the data from the subnet to the internet. That requires a router and a gateway on top of the Wi-Fi connection. Full-blown computers in their own right, routers and gateways will offset the low costs of any network node.
Cellular networks present their own cost issues. Mobile M2M networks such as trucking fleet management systems need the near universal coverage provided by cellular networks. The assets - in this case, the trucks - have GPS receivers tied to wireless modules that call home through a cellular network. While the cost of the wireless module is declining rapidly, the cost of using the network is still relatively high due to the telecom companies' pricing models that charge for connections or minutes when an M2M application often needs to send only a few bytes of data.
So trade-offs need to be considered for both today's technologies and the potential impact of emerging technologies such as Wimax on massive M2M deployments.
AMR: M2M's ticket to stardom
Automated meter reading (AMR) will be M2M's leading application according to Therese Cory, an analyst with Juniper Research specializing in IT and telecommunications.
"The utility industry is a prime example of how, by networking and remotely monitoring machines, data can be analyzed and collective behavior understood in new ways," Cory says.
"For example a real-time unified view of how power is used will help safeguard this increasingly rare resource. And what can be measured can be controlled, and ultimately optimized."
To date however, growth has been restrained. M2M projects are notoriously long-term investments taking years to initiate, due to large budgets, the length of the decision-making process, and time needed to put together a custom solution from several supply chain players. Despite these hurdles, Cory insists the signs are positive.
Enel, the Italian energy company, has completed a $4.5 billion installation of 30 million smart electricity meters. The meters send their data through the existing electricity grid to the nearest substation. The substations then pass the data on to the centralized control rooms through GSM.
Others in Europe initiating AMR programs include Sweden, the Netherlands and the UK. Closer to home Australia has started a program in Victoria. Heavy summer time use of air conditioners was causing extra investment in low-use power plants. The Essential Services Commission of Victoria pushed for legislation, resulting in a program that will see about one million smart meters installed by 2013.
Malaysia's pilot projects used GPRS and mobile RF systems, but proved to be very expensive. Last year, Malaysia's largest electricity utility announced a contract to deploy AMR meters to commercial and industrial customers. The system will use GSM.
Frost & Sullivan reports that other Asian countries are looking at RF technology utilizing handheld, mobile, or fixed networks for their AMR systems. Thailand's Provincial Electricity Authority is following up on a 30,000-meter system. Indonesian and Vietnamese utilities are also testing pilot projects.
An IP address for every machine
Futurists envisioned M2M networks as having all their assets connected directly to the Internet. Set aside the cost issue of each asset needing a radio and also the protocol stack and hardware for TCP/IP, much less the problem of assigning IP addresses for all these assets and then providing DNS or some other means to find and use these assets. Today's reality is that there is not enough available addresses to go around.
IPv4, the current addressing scheme, allows for about a maximum of 4.3 billion addresses. However, this has proved to be not enough as more and more devices are hooked up to the Internet. Currently ISPs are using a work-around, Network Address Translation, while everyone waits for the new addressing scheme - IPv6 - to be implemented throughout the system.
To make sure that the net won't run out of addresses in the near future, IPv6 allows for about 3.4×1038 addresses. That's enough to place about 660,000,000,000,000,000,000,000 devices on each square meter of the earth's surface.
That should be enough for M2M.
- Chris Everett