M2M Middleware Platforms

Sean Horan of AT&T recently wrote a very good blog entry on “4 Reasons to Justify M2M Middleware Solutions.” Basically, Mr. Horan argues that M2M middleware makes it faster and easier to develop a broader range of more functional M2M applications. He especially sees M2M middleware helping large enterprises implement diverse M2M initiatives across their business units.

The four major benefits of using M2M middleware that Mr. Horan points out are:

·         Support for Multiple M2M Device types
There is a large and growing variety of M2M devices that an application or set of applications may need to support. Each of these devices comes with their own protocols, particularly for device management, and middleware helps insulate application developers from these differences by making their applications device agnostic.

·         Reduce M2M application development costs and standardize deployment
M2M middleware provides a number of M2M application and communication functions, as well as a framework to simplify the development of the application’s custom logic. M2M middleware also provides the architectural underpinnings to enable users to scale and manage M2M applications easily.

·         Integration enablement
The data collected from remote M2M devices is useful only if processed, analyzed and used in central applications. This is particularly true for larger enterprises in which the M2M data is often used to enhance the operation of existing enterprise applications. M2M middleware makes it easier to interface M2M data or applications with general enterprise applications.

·         Support Multiple Connectivity optionsBroadly speaking, M2M applications need to communicate across a broad range of wireless networks in addition to cellular – WiFi LANs, satellite, and low power wireless technologies. Applications also sometimes need to accept connections over the Internet. M2M middleware helps applications to be network agnostic and well as device agnostic.

Without a doubt, M2M middleware platforms enable you to develop and deploy M2M applications more easily and quickly. They can help deliver the four important benefits that Mr. Horan describes.

But there are two architectural approaches to M2M middleware that differ significantly in their approach to trying to deliver these benefits. They each have their pros and cons. It is advantageous to use an M2M middleware platform in developing your application, but it is critical to pick the middleware architecture that best meets your application's requirements and environment.

One type of M2M middleware platform follows the paradigm of classic information systems client/server middleware, in which one half of the middleware is implemented on the central application server, and the other half (the "agent") is implemented on the remote device. This approach can deliver all of the benefits Mr. Horan describes, and it is especially effective for migrating the rules engine, exception handling, and sophisticated diagnostics to the network "edge." This is particularly beneficial for applications that operate over networks that charge for usage, like cellular networks.

The downside with this approach is that EVERY device manufacturer involved in the application must implement the agent in their firmware. This may not be an issue if you are developing the hardware in-house, or only buying a single device from a manufacturer with whom you have leverage. This is more challenging if you are trying to get the agent implemented by the manufacturer of the vehicle-tracking device, the retail kiosk manufacturer, the vending machine manufacturer (all models), etc. Also, these agents are not lightweight, which is often an issue with many embedded devices. Implementing an agent in the full computer that is built into an MRI machine is easy; implementing it on the 8-bit microcontroller in a tank level monitor, not so easy.

The other major type of M2M middleware platform centralizes the device translators on the server and allows the many multiple M2M device types to communicate to the central server in their unaltered, native mode. This makes it easier to incorporate devices from many different manufacturers, and to handle the manufacturers' upgrades to hardware and firmware. The manufacturer does need to share their devices' communications protocols so that the appropriate translator can be created, but this is usually easier than getting the manufacturer to incorporate licensed agent firmware into their product. Otherwise the equally powerful server component of this middleware platform delivers the four major benefits that Mr. Horan profiled.

The downside of this approach is that functionality is not standardized across devices, particularly for device management and diagnostics. It is also more difficult to optimize application performance and wireless costs when incorporating unaltered, inexpensive, off-the-shelf M2M devices.

Either one of these major types of M2M middleware platforms can deliver the major benefits that middleware provides -- multiple device support, easier application development, integration enablement, and multiple connectivity options. But the specific difficulty and cost of developing, deploying and operating your application can differ markedly between the two architectures. You should carefully consider how your application requirements match up against the specific capabilities of the two approaches to select the best middleware for you.

What Do Carriers Really Think of M2M?

There has been some discussion in the trade press and in analysts’ reports that cellular carriers are starting to focus on M2M as their next growth opportunity. Many of these carriers in both the United States and Europe have established or greatly expanded their M2M-focused business units to try to grow and capture this emerging market for cellular devices and service.

For many reasons, the M2M business should be very attractive to carriers. First, the potential market size is very large – there are many more machines and devices in the world than there are people, and carriers like large markets. Second, the M2M business is largely a wholesale business, with some other company selling the product or service which incorporates cellular communications, so the carrier does not take on the significant expense of marketing and sales. Whether it is a tablet computer, an e-book reader, a home health diagnostic unit, or an oil pipeline monitoring device, the manufacturer of the product sells the cellular service as part of the overall product. Third, the M2M market is developing without the use of upfront equipment subsidies, as are common in the US cellular handset market, so that the carriers do not face a large upfront investment to add each new M2M “subscriber.”

There are other characteristics which should make the M2M business economically attractive to carriers.  Because this is largely a wholesale business, the acquisition cost of each individual subscriber is very low. In addition, outside of the consumer data appliance category, the average service life of the devices using cellular communications is very long, sometimes as much as a decade.  This means that the churn of M2M devices is very low. Except for consumer appliances, most M2M applications tend to be tightly defined and change little over time. Once they are working, they tend to simply keep working, so carrier support costs are relatively low per unit. In fact, most direct support is actually provided by the product manufacturer. Finally, the network usage of M2M devices tends to be relatively low, predictable and stable, so these applications do not place unforeseen demands on network capacity (other than simply adding more devices to the network).

But carriers are actually somewhat ambivalent about the M2M business. The problem for them is that M2M devices usually provide a much lower monthly Average Revenue Per Unit (ARPU) than do traditional cellular subscribers. In fact, M2M ARPUs are often a full order of magnitude lower than standard consumer handset ARPUs.

Cellular carriers are public companies whose operational performance is evaluated on three numbers – subscriber count, average monthly revenue per unit, and churn (i.e., the percentage of customers who drop cellular service with that carrier during the month). While the stock price of cellular carriers does partly reflect the carriers’ recently reported quarterly financial performance, investors also attempt to gauge the carriers’ future financial performance by tracking these three operational metrics. These three numbers impact an investor’s view of the carrier’s future prospects, and investors incorporate this view of the future into the stock price.

In terms of these investor metrics, then, a significant increase in M2M subscriptions will cause the subscriber count to increase, but will also cause the ARPU metric to decline. The greater the number of M2M subscriptions added, the more dramatic will be the impact on the carriers’ overall metrics. But when investors see a carrier’s overall ARPU numbers declining, they can not easily tell whether that is caused by a growth in M2M subscriptions, or whether it is driven by general price erosion in the consumer handset business, from customers shifting from higher cost plans to lower cost plans that better fit their usage, or from a shift in customers from traditional post-paid cellular plans to prepaid plans. All of the other possible explanations for declining ARPU are generally viewed as “bad” from the investors’ perspective, since they threaten to reduce the carrier’s future profitability from declining revenue with no corresponding decline in costs.

The reason that carriers are actually ambivalent about the M2M business, and remain uncommitted to really growing it, is because the faster this business grows, the less attractive the carrier looks to investors. Even though an M2M business might be extremely profitable as a stand-alone business (even after allocating its share of the full cost of the network capacity that it consumes), it is impossible to convey this fact to investors using the summary financial and operating reports that carriers disclose publically and that investors use to evaluate carriers. So growing an M2M business will make a carrier’s stock price go down – at least in the near term. Like most public companies, carriers focus on short term stock price performance as their primary guide for managing their business. Most of the carrier palaver about supporting M2M is really about high end consumer appliance subscriptions, not about the average low ARPU M2M applications.

Reality Check – The Cost of M2M Data Services

The M2M industry is still relatively small and new, and consequently many application developers are planning and developing their first M2M application. Many of these developers’ only prior experience with the cost and capabilities of cellular service is through their personal experience using consumer cellular appliances, particularly smartphones, tablets and data cards for laptops. In particular, their frame of reference for the cost of cellular data comes from its cost with these devices. Their assumption is that if they can develop an M2M application that uses a half or a tenth or less of the data of these consumer appliances, the cellular cost of will also be proportionately lower.

And in this assumption they will be wildly wrong.

The cost of cellular data service on personal devices has declined significantly over the past five years as the popularity of intelligent connected devices has increased, and as the range of available appliances has also grown. Smartphones generally did not exist five years ago, and cellular data services were largely confined to PC cards that were inserted into laptops. Phones had Internet access back then, but it was cumbersome and lightly used. Now smartphones with downloaded applets make up over 40% of US mobile subscribers in late 2011. On top of that, tablets have experienced explosive growth in adoption, along with e-book readers, digital picture frames and other graphical devices, many of which have cellular data connections.

It is easy to see what pricing conventional wireless data users expect. With smartphones, it is tricky to isolate just the cost of the data plan from the overall cost of the service since the monthly charges for smartphones include voice and text messaging services, along with amortization of the subsidized purchase price. The data plan price on a smartphone is an incremental price. With that caveat, the minimum incremental data plan price for an iPhone, at the end of 2011, is $15 for a 2 GB data plan, which works out to $0.0075/MB per month. More “pure” data plan pricing offerings exist for the iPad. Monthly data plan prices on the iPad range from $15 for 250 MB to $25 for 2 GB, which calculate to monthly data costs of $0.06/MB to $0.0125/MB respectively. The iPad pricing is 2x to 4x the incremental data pricing on an iPhone, and is the better example of wireless data pricing on a consumer data-only device.

The situation in M2M applications is markedly different. M2M applications use significantly less data per month than do consumer devices. There are a number of M2M applications that require miniscule data traffic, less than 100 KB per month. Even many verbose, “high volume” M2M applications require no more than 1 MB to 5 MB of data monthly. In general, M2M applications use at least two orders of magnitude (i.e., 100 times) less data than are used by consumer appliances.

Given their lower data usage, you would expect the monthly service cost for an M2M device to be lower -- and it is, but not as low as the consumer device benchmark would suggest. Typical M2M pricing for a device using a miniscule 100 KB of data monthly is $2 to $3. That is the equivalent of $20-$30 per megabyte. Higher data volume M2M applications, which consume 1 MB to 5 MB monthly, also have relatively low monthly charges of $5 to $8 respectively. Those monthly charges, however, are the equivalent of $5/MB to $2/MB.

Clearly if an M2M application only needs to use 100 KB or 5 MB of data service monthly, then M2M pricing from carriers is less expensive than putting the device on a standard consumer data plan. But the price for M2M data is extremely expensive, by comparison, when measured on a per megabyte basis. In fact, the price per megabyte of data service for an M2M device is two to three orders of magnitude (i.e., 100 to 1,000 times) more expensive than the price charged for data for consumer appliances. An order of magnitude difference is a glaring difference – two or three orders of magnitude is huge. This pricing difference is one of the reasons why new M2M application developers are shocked when they calculate how much their application is going to cost each month, given how little data service they are using.

Both sets of devices – consumer appliances and M2M devices – are using the exact same cellular network with the exact same data communications facilities. So why do carriers appear to charge 100 times more for the capacity used by M2M devices compared to that used by consumer appliances?

One major difference between the two sets of devices is breakage. Breakage reflects the fact that subscribers almost never use their full data plan allowance each month. Because cellular users of all types – both businesses and consumers – hate to pay the overcharges incurred when they go over their plan allowance, cellular users invariably sign up for a plan that provides more data than they need. The difference between the allowed usage in the plan and the user’s actual data usage is called breakage. An iPad subscriber on a 2 GB plan may only use 500 MB of data in a month. The unused 1.5 GB allowance is lost, which makes the iPad user’s actual data cost that month $0.05/MB, not the $0.0125 that optimal, full usage of the data plan allowance would provide. (As an aside, some carriers instituted “rollover minutes” to reduce breakage on cellular phone plans as an effective price reduction. There are not currently any “rollover megabyte” plans in the industry).

Breakage differs significantly between M2M devices and consumer appliances. Because consumers do not understand their data usage very well, and because carriers only provide a few data plans of significantly different sizes, most consumers pay for a much larger data allowance than they need. Average breakage on consumer data plans is probably over 50%, and may be as high as 80%. By contrast, M2M application developers usually know exactly how much data their devices will consume each month since the application behavior is tightly defined and controlled. Also, M2M data services are usually available in a large number of pricing tiers so that the application developer can select more closely the appropriate data allowance for the application. And finally, some M2M pricing plans are available as “pooled” plans across a large number of devices, in which the under usage one month by one device can be used to offset the over usage that month by another device, thus largely avoiding overcharges. Consequently, breakage on M2M data plans is much less than 50%, and probably closer to 20% on average.

The difference in breakage then may mean that M2M data is only 75 times more expensive than consumer appliance data, rather than 100 times. It does not begin to explain the wide variance in the cost of cellular network data capacity implied by the two sets of pricing.

Another consideration is that M2M applications place a proportionately larger load on the signaling capacity of the cellular network than do other data applications. Signaling is used on the cellular network to start and stop data communications sessions. Carriers only charge for the data contents of the session, not for the signaling transactions themselves. An M2M device using 1 MB of data over the month may have the same number of data sessions as a consumer appliance (say, 1000 separate sessions). The carrier is collecting more money from the consumer appliance’s higher data traffic to cover the “cost” of the signaling than from the M2M device, which is only consuming a relatively small amount of billable data for the same signaling. Consequently, carriers are probably charging more for the measurable, billable M2M data to help cover the signaling overhead costs incurred by M2M applications.

The fact, though, is that neither of those reasons, nor any other cost driven explanation can justify the extravagantly higher price that carriers charge for M2M data compared to any other data service. M2M applications do not incur two orders of magnitude more load on the cellular network per megabyte of data communicated (or at least, per megabyte of data allowed).

The explanation has to be that carriers are value pricing M2M data. They are grabbing as much revenue as the market can bear. Because the M2M market is still relatively small, because M2M applications give the lowest monthly revenue per device that carriers see, and because carriers do not believe that lower pricing will generate significant new revenues quickly, they feel no need to compete in the M2M market based on price. To some extent, carriers still view true M2M applications as more of a nuisance business that they reluctantly agree to accommodate. A carrier’s business is not going to be significantly impacted if high pricing keeps a million $5 per month M2M devices from being developed and deployed on their networks (and by the way, there is not a single 1 million device M2M application deployed in the US today – most are much smaller than that). Carriers reserve their most aggressive data pricing for those devices that look more like smartphones – tablets, e-book readers, etc. – which individually provide more revenue per month or which can be sold in very large volume to one entity (e.g., Amazon).

Carrier value pricing of M2M data will eventually give way to more aggressive price competition as the M2M market grows large enough to be worth chasing. Until then, M2M application developers will have to closely watch the ROI of their solution and design their applications to minimize on-going data costs.