Enterprise 2.0 Session: Machine-to-machine - opening for business
This is a summary of the first half of the Enterprise 2.0 - New Opportunities in Wholesale and Enterprise Services session at the 9th Telco 2.0 Executive Brainstorm
The M2M session produced a buzz at the 9th Executive Brainstorm, held in London on the 28-29th of April. This was partly down to the excellent quality of the speakers and the subsequent panel discussion. However, there was also a strong belief that M2M's time is finally coming and a determination to learn the lessons from the past and from elsewhere, more specifically, from the Internet.
The panellists (Phil Laidler, Telco 2.0 Director of Consulting; Ken Figueredo, Principal, Ventura; Marie Austenaa, VP Strategy & Products, Telenor Objects; Jonas Nordstrom, Head, M2M Business Development, Ericsson; Alan Beveridge, Director, Sales and Business Development, Jasper Wireless; Kevin Johnson, Director, Embedded Systems, Intel; David McNierney, VP, SAP) all demonstrated a clear understanding of the importance of M2M achieving future growth through facilitating innovation and of the challenges this line of business presents - notably extreme customer diversity, a mismatch between traditional product development processes and many markets, and certain key technical issues, such as power consumption and remote device management.
There was consensus on the need for horizontal platforms (albeit different visions of the how the ecosystems might work in practice) , a two-sided platform business model, and engagement with external innovators. Telenor presented an impressive new venture launched under a year ago - Telenor Objects - an M2M platform implemented as open-source software.
What we saw was not so much the Industry rethinking M2M, as the industry getting on with redoing M2M. Visionary concepts perhaps, but very practical steps too. Heady stuff indeed.
A key theme at last week's Telco 2.0 event was the way in which operators and vendors are beginning to make the changes they need to realise the potential of the machine-to-machine market. This includes machine-to-just-about-any-device, the Internet of Things or Embedded Mobile. But what are the key enabling ecosystems?
The industry has already been here a few times before; there has been a lot of confident talk about 400% mobile penetration in the past and rather little delivery. But what we heard at Telco 2.0 suggests that at least some players are now approaching the opportunity in the right way.
Telco 2.0's director of consulting, Phil Laidler, who opened the Enterprise 2.0 session, remarked that Telco 2.0 in the enterprise is all about automating interactions, which can be machine-to-machine, application-to-person (A2P), or person-to-organisation (P2O). M2M systems seek to facilitate useful interaction between machines, or between machines and people monitoring them.
P2O systems are all about interactions between individuals and organisations - the most obvious examples being marketing, sales, order management, customer relationship management and customer care. The same ideas apply to internal enterprise applications where the people on the phone are employees or suppliers. To date, M2M has typically involved narrowband data, messaging, and occasionally video in surveillance or safety-critical monitoring applications. P20 has more often involved voice, messaging, email and increasingly the Web.
Looking at M2M in this way challenges some assumptions. Traditionally, "Industrial" M2M has been an extension of SCADA (Surveillance, Control, and Data Acquisition) systems, primarily concerned with collecting data from remote sites and delivering them to a central control point. More rarely, the control centre could also send an instruction back to the remote site to take action. These systems tend to be built from scratch to meet individual enterprises' requirements. They also tend to adopt proprietary (or at best, sector-specific) technology, and to have a highly-centralised architecture. They are often difficult to integrate with other systems, and not interoperable with systems in other companies.
From an interactions perspective, though, it might be very valuable to have the machines operating with greater autonomy and interacting directly with one another - this would permit much faster reaction to events, would cut the volumes of data that need to be processed centrally in real time, and would permit the system to be robust against loss of communications. If the devices in the field are smart enough to apply the analytical rules themselves and issue the commands to maintain stability, it becomes significantly less worrying if a network link goes down.
It has been very difficult, to date, for other businesses in a value network to benefit from a company's M2M data; for example, the players at an oil terminal might include multiple oil companies, contractors, suppliers, the port authority, shipping lines, pipeline operators, a ministry of oil, and the customs service. All of these have a need to access data relating to the flow of oil through the pipelines; if they could all get data from the instrumentation system, it would solve many of their problems.
Another issue with M2M, is that you might solve one set of interactions by creating a huge new interactions problem. So, you have deployed hundreds or even thousands of sensors and control units and networked them. Some of them are built into long-lived pieces of equipment - wind turbines, trucks, and the like. Some of them are safety-critical, or installed in places that are difficult to access. You have then set up the data-processing backend. Now, how do you manage the devices? Without good solutions for this, the project has created a large number of costly maintenance calls that cannot themselves be easily automated.
Also, as all the speakers pointed out, one of the biggest challenges in M2M is that there are many companies out there, with immensely diverse business processes, interacting in complicated ways. The very promise of M2M - a market as big as the entire industrial economy - is also the biggest obstacle. Only the biggest customers are worth doing a tailored implementation for, as nearly all M2M systems are bespoke. As Phil Laidler pointed out, - simple applications for big customers tend to get done, simple applications for very small customers and the more complex, les stable interactions always end up being carried out by human beings. That leaves a great swathe of processes undertaken by mid-sized operating units (often part of much larger private or public sector organisations) who are left underserved.
The good news, is that if you can identify the common elements in the set of interactions associated with each industry vertical, you should be able to horizontalise them - to create a set of building blocks that apply to all the M2M verticals. If you then do this in an open, inter-operable manner, you start to create the infrastructure for the Internet of Things.
Embedded is horizontal
Ventura Consulting principal Ken Figueredo was the first speaker to touch on this, pointing out that there was a crucial strategic choice between creating a horizontal platform and limiting your activities to a few key verticals. Ken explained that the GSMA concept of Embedded Mobile was much broader than the typical industrial M2M concept, and that this meant both more opportunity and even more complexity.
One way to deal with the complexity, as well as developing horizontal capabilities of general applicability, is to look for partners who can provide channels to market. Figueredo cited Vodacom's partnership with Sierra Wireless as an example - as a major vendor of embedded mobile equipment, they already have a customer base and channels of their own, which allows Vodacom to reach the enterprise buyers who are going to make the decisions. Eventually, he said, it would be critical to identify the key "value intersections", pointing out that M2M-focused MVNO Jasper Wireless, software vendor SAP, and Telenor were all targeting different ones, but that they were all betting that one particular role in the ecosystem would be strategically crucial.
Telcos' role in the value chain: supporting innovation
Kevin Johnson, Intel's director of embedded systems, arrived brandishing an Intel Atom-based system on a chip. He pointed to the vast numbers of possible sub-segments in the market, and asked which layer in the technology would concentrate the most value. The uppermost layer was display - visualisation dashboards, user interfaces and tools. Next down was analytics and data-mining. On the bottom layer, were the sensors, gateways, and controls themselves. He argued that both the value and the problems were concentrated in this hardware layer, and that horizontalisation was critical in order to scale-up the business.
Kevin stated that "the innovators are your best friends". He said that operators needed to find a role in the value chain that supported innovators. For example, creating a building-automation system involved a range of different actors, including the chip and RF-chain vendors, the device assemblers, the operating system vendors, the network operators, and the application developers. He argued that the role of the developers was the only one in the value chain where all the other domains overlapped - hence it was the critical node. They had to deal with all the technical partners, plus the end user.
Kevin estimated that in this example, developers would collect about 80% of the total contract value (on a €1m contract), with about €1,500 a month of revenue accruing to the network operator. In order to "make it happen", he said, there were two critical elements that had to be addressed:
- "Defragmentation", in which the proliferation of proprietary platforms had to be reversed, the CPU and radio elements better integrated, and software APIs standardised
- R&D. The two main R&D goals should be to cut power requirements radically, and to solve device-management problems.
Telcos would need to work on network optimisation, but that paled in comparison with the prospect of 100 million support calls. He asked why major billing systems vendors weren't doing more to fulfil the BSS-OSS requirements, and pointed out that getting the tariffs for M2M traffic right needed experimentation. Summing up, he said that it was crucial to define an ecosystem for innovators, that the industry needed to speed up the device approval process through mechanisms such as Verizon ODI, and that efficiency required innovators, scale, and devices.
Looking for M2M hot-spots
Jonas Nordstrom, head of M2M business development at Ericsson, presented the results of a market research exercise conducted with key actors in the industry. He used the classical S-curve and hype cycle models to predict that there would be a major wave of deployment in the next six years, and pointed out that broadband was still in the phase where most value accrues to the original players. The interviewees were asked to identify the verticals they considered most disruptive, based on their market impact and the speed of their expected uptake. The top five were identified as Connected Car, Smart Grid, Fleet Control, Mobile Surveillance, and Medical Monitoring.
In the healthcare field, the interviewees felt that data management, automated diagnostics, and virtual and augmented reality were the most important in the foreseeable future. It would be vital to get the buy-in of the clinicians and the pharmaceutical industry.
In automotive - which will clearly be a critical sector if the interviewees are right - the research pointed to electric vehicle recharging, maintenance, and presence in the BRIC+5 countries (the BRICs and Iran, Indonesia, Argentina, Mexico, and Thailand) as important. Jonas described another research project being carried out by Daimler-Benz and MAN into "Cooperative Cars". This would have cars communicate with each to capture and disseminate traffic information - as each GSM call generates two location updates, operators might well have a role here.
The next step, according to Jonas were was to identify the key horizontal capabilities needed. Jonas ended by confirming that the main barrier to M2M success was winning internal support within our industry.
Telenor Objects - Leading the way
We had heard a good deal about horizontal capabilities and platforms by this point. The next speaker, Marie Austenaa, is VP Strategy & Products for Telenor Objects, their new dedicated M2M division. It turned out that Telenor Objects is essentially an effort to implement the sort of horizontal, multi-tenant platform for M2M applications that others have been advocating.
Marie Austenaa remarked that Norway has a lot of space and some very complex industrial infrastructure, but not many people, so Telenor had long been asked to implement vertical M2M solutions. Past customers had ranged from oil and gas to fish farms, and one system had even been used to monitor a flock of sheep. Over time, Telenor had come to realise that the same functions were being implemented again and again in every M2M system they built.
The verticals Telenor worked with had many things in common. They had to collect data, send back commands, carry out data processing according to user-specified rules, and manage the device fleet. Telenor had also begun to realise the limitations of classical M2M; "it wouldn't work on the Internet of things". Far too much of these systems consisted of closed-source, monolithic, standalone applications. Little code was reused, and much of the hardware was task-specific; sometimes it was necessary to junk the hardware in order to make changes in the software.
Their answer was to use this to launch a horizontal approach to M2M, based on the question "What things do all the verticals do?" As a result, they launched two new companies within Telenor:
- Connexion, to provide vertical systems and connectivity, and:
- Objects, to build the horizontal platform.
The design of Telenor Objects' software platform began with a layering concept. Telenor would provide a middleware layer to help different kinds of devices, networks, and applications to interoperate. This would provide open interfaces for data capture, data pre-processing, device management, and information exchange with other systems. This last feature is critical; for example, transport companies frequently borrow and charter vehicles from each other. What is most striking about Telenor's approach, is that it looks very different from the traditional Telco 1.0 mindset of vertical integration and control.
Telenor Objects is based on five key elements - the brand and channels to market, the technology, the partners, a managed-service business model drawing on the telco heritage, and open-source software. The complete system provides a range of reuseable capabilities, including a GUI for users, developer APIs, and a device library. The core of the whole project is the software platform, Shepherd, which implements a secure message-exchange and device management system. Shepherd has generic software libraries for each group of devices by function in the device library. Telenor has released the source code for Shepherd as open-source software through SourceForge.net; at least one delegate actually grabbed the codebase to his laptop during the session.
An interesting feature of Shepherd is that it provides for distributed data processing in the network of devices - users' business rules can be injected into the system, so that control can be delegated down to the device gateways, implementing the sort of distribution we mentioned above.
Finally, she said, it had been necessary to take a strategic decision to shape the market. A pure market approach had led to a succession of suboptimal vertical silos. In essence, though, the business proposition of Objects was a subscription-based service under which Telenor managed devices and customers paid them money.
Lessons from the AppStore
David McNierney, a VP of enterprise software vendor SAP, had the hard act to follow. He introduced the M2M market as being potentially huge but - here it comes again - immensely diverse and complex. He explained that SAP's competence in providing ERP, workflow, and accounting/audit solutions for complex business processes was cross-applicable. In his view, the biggest customers were likely to be logistics, transport, and medicine, and he wondered how well placed telcos really were to serve them.
He referred to lessons from the Apple App Store, pointing out that this was an effort to serve a hyper-diverse market of individual customers and individual developers. Partnering, monetising, and analytics dashboards had been crucial competences. It would be very important to negotiate the terms of the partnerships involved in M2M; a special issue would be the degree to which the terms of contracts "cascaded" to the third-party developer's own third parties. M2M applications would also be very demanding in terms of filtering the streams of data they generated.
Finally, M2M startups might struggle to scale up, and therefore there was a need for "strategic relationships" with a triangle of telcos, hardware vendors, and companies like SAP, IBM, BT Global Services, etc.
Getting the machine into gear
During the panel session that followed, perhaps the most significant statement was from Telenor again. "A huge standardisation issue" had been identified in healthcare - furthermore, there was a huge culture gap between the medics, pharmaceuticals, and the tech industry. Marie Austenaa replied that "if there was a standard it would be so much easier, so let's do one!"