The big players may be fighting to drive and control the software platforms for non-phone devices, but there is an even more open playing field when it comes to an even bigger prize ‘ providing the framework that will underpin the internet of things (IoT).
This will be multi-faceted, of course, but at the current stage, there are key roles for which to audition ‘ cut-down operating system, applications framework and store, peer-to-peer protocols, the broader intelligence to coordinate the billions of ‘things’ and process their data. There will be big debates over whether software platforms and protocols developed for human communications, or for the enterprise data center, can be shoehorned into a world of vast numbers of automated, low-resource gadgets and sensors. So start-ups and universities are proposing new approaches optimized for this environment, but the majors of wireless and data platforms will be determined to adapt their offerings for this huge potential opportunity.
IBM is hard at work at protocol and cloud level; Intel is making a smart, if largely under-the-radar, play based around its hugely important Wind River acquisition. And in the past weeks, two giants of the mobile arena have been talking about their own contributions ‘ Samsung describing SAMI, a kind of Siri for ‘things’; and Qualcomm evolving its AllJoyn peer-to-peer technology into a full IoT platform.
Samsung has poached Luc Julia, the former director of Apple’s Siri system, the artificial intelligence-based personal assistant which made its debut with the iPhone 4S. Siri has become a household name because of Apple’s marketing impact, but other companies are working on more sophisticated AI engines to power future mobile services and behavior ‘ in particular Google, which desperately needs to keep control of the search experience, is adding AI into its platforms via Google Now.
Samsung, increasingly diverging from Google in the Android experience and creating its own platform, has also taken a strong interest in AI-driven mobile apps, but it is taking this a big step further and applying its technology, dubbed SAMI (Samsung Architecture for Multimodal Interactions) to IoT as well. In that respect, it hopes to steal a march on both Google and Apple, as well as the giants of AI research such as IBM (which, interestingly, opened up its famous Watson AI search engine to mainstream developers this week by releasing an API).
Julia demonstrated SAMI at last week’s MEMS Executive Congress in San Francisco, close to his base as innovation fellow at Samsung’s Open Innovation Center in Menlo Park, Silicon Valley. This center is vital to Samsung’s efforts to improve its software capabilities and complement its efficient hardware design with advanced R&D, important hedges against the probably inevitable peaking of its dominance of smartphones. SAMI is one of its most important projects because, while drawing on some of the principles of Siri, which Julia helped create at Apple, it goes far beyond that. In particular, it aggregates data from all kinds of sensors and other IoT devices, from any supplier, in the cloud and then performs deep analytics.
This is nothing to do with Samsung developing IoT hardware and everything to do with creating an ‘open’ platform which attracts massive developer and partner support, but which the vendor still effectively controls ‘ the same trick Google pulled off with Android. SAMI will initially be available for Samsung’s ecosystem partners to harness in their products and apps, and in future is likely to be opened up to the broader community via APIs and even open source code.
“IoT is all about sensing and collecting the data, and then transporting it to the cloud] Once it’s there, it’s about massaging it and sending it back to the user,” Julia told the congress. This is not just about handling ‘big data’, but analyzing that so intelligently that the system can send back ‘very small data’ ‘ precise and relevant results ‘ to the end user, the process automated with sophisticated AI and machine learning techniques.
SAMI ‘ currently an internal project of six months’ standing – aggregates data from sensors, filters and augments that and then stores it in a normalized format before applying analytics. Results can be accessed via APIs or sent directly to an end device such as a handset or smart watch. The aim is to address the problem of the huge variety of data formats that are present in IoT ‘ if Samsung can power the platform that unifies all that, it will achieve the sort of influence over the ecosystem that IBM has enjoyed in some enterprise and vertical data areas. Julia, EETimes reports, is concerned with breaking down data siloes and allowing smart apps to access any kind of data, including structured or unstructured, using common methods.
“We tell start-ups just to give us their data in the format they already have, and what we do is create a normalized a way of accessing it,” he said. “Security is also very important; we need to control who has access.”
SAMI is part of wider IoT efforts going on in many of Samsung’s 27 Open Innovation Centers (OICs) worldwide, which will drive its strategy of investing heavily in software capabilities to broaden its business model ‘ as outlined by top executives at the firm’s first ever analyst day this month (see Wireless Watch November 7 2013). The Menlo Park Center also houses a $100m accelerator fund, which has a heavy focus on software frameworks and IoT, and the OIC says it is exploring possible projects with 48 companies in this area. Topics include developing sensors and actuators for smart cars, smart homes, health applications and wearables, as well as the back end systems for analytics.
Every company has a prediction for the number of IoT devices which will be connected by the end of the decade ‘ Julia’s estimate is that there are 20bn IoT gadgets today which could be linked up, and that will grow to 1.5 trillion sensors by 2020. His figure is far higher than those from Ericsson and others because he is counting the multiple sensors which will support increasingly multitasking smartphones – he envisages sensors ‘exploding like shrapnel’ in handsets and M2M products, and becoming embedded in wearable devices all over the body.
Indeed, he presented to the congress wearing six different sensors himself including a Basis smart watch, Fitbit activity monitor and Vitalconnect ECG heart monitor. The data from these gadgets was combined with weight measurements from a Withings scale and food intake data from a HapiLabs smart fork. Julia verbally asked SAMI “How am I doing?” and was presented with a dashboard summarizing the various findings and offering advice (such as how much weight to lose).
Meanwhile, Qualcomm’s IoT efforts have been developing for years, with efforts such as an experimental apps store and ultra-low power modems. Perhaps the most strategic development is AllJoyn, the proximal peer-to-peer networking technology that started as a convenient way for nearby handsets to communicate, but is being evolved into a fully fledged platform for all kinds of connected devices.
The new IoT-oriented iteration of AllJoyn will be shown off at January’s Consumer Electronics Show (CES) and the chip giant is eager to open the framework up to partners in the bid to create a de facto standard. Sy Choudhury, senior director of product management for the platform, was talking it up at a Qualcomm workshop in San Francisco, and told Mobile World it had evolved from a ‘smartphone app to smartphone app’ technology to be a communications system for other smart devices, and from 2014, for IoT appliances too.
Like Intel, Cisco and others, Qualcomm talks in sweeping terms about IoT, but in reality knows that, for commercial purposes, it needs to select a few specific targets first and build up its base step-by-step. Its choices reflect the positioning of its core chips ‘ like Snapdragon, AllJoyn will address home multimedia applications first. Its first real IoT focus, for next year, is home automation, while it is carrying out R&D in areas such as automotive, but without commercial plans for the coming year.
AllJoyn is being constantly augmented, and most recently Qualcomm has added a common way to perform certain key tasks, such as sending and receiving notifications across devices, to enable multivendor interoperability. The framework’s aim is to establish ‘a common lingua on top of the base peer-to-peer connectivity that the software enables’ ‘ enabling an open ecosystem of IoT devices and apps, but with Qualcomm’s technology acting as the glue. As with SAMI, this is all about a company embedding itself into the fabric of IoT so that no individual vendor or service provider can avoid its underlying technology ‘ a source of power and partnerships as much as direct revenue.
Of course, this only becomes possible if other power players ‘ and innovators ‘ see sufficient merit in the platform to outweigh the possible downsides of being tied into another firm’s underlying technology. So openness, at least at one level, is essential, and Choudhury claims ‘very large companies’ are incorporating AllJoyn into devices. In the interview, he said these firms are saying ‘it’s great, and its licensed under permissive open source, but how do I know that Qualcomm won’t change its mind about that? So we want some sort of a standards body’.
Qualcomm is taking that hoary old route to technology power ‘ claiming its platform does not fit well into a formal standards body, and so setting up its own ‘industry alliance’. This will happen within the next 2-3 months, the firm says, since it sees no obvious place for ‘connectivity middleware’ in existing bodies. The group, like so many alliances of its kind, will provide independent governance of the code base, making it more neutral and so encouraging outside contributions. And it will also act as a testing, certification and branding authority. Of course, the relationship is likely to be similar to that of the Android Open Handset Alliance with Google ‘ Qualcomm providing the bulk of the R&D and the fundamental code and so keeping control. There’s a chance, if AllJoyn does prove to be sufficiently valuable to the IoT effort, that another major will hijack it as IBM tended to do when Sun owned Java ‘ but Qualcomm is a far wilier player than Sun ever was when it comes to industry bodies.
However, this is a very new approach for a company whose business model rests so heavily on licensing the technologies it develops. Qualcomm, ever the chameleon when it has to be, has accepted the dominance of the open source (or at least the vendor-driven open source) model in the new software world and is determined to play ‘ however difficult the internal debates must have been. The company set up a separate open source subsidiary, Qualcomm Innovation Center (QuIC) back in 2009 and it was this unit which announced AllJoyn as a system for ad hoc communications in early 2011. The division was set up with a brief to ‘optimize open source software with Qualcomm technology’ and to ‘use open software to spur the development of advanced services’, allowing the company to innovate in areas outside its traditional business.
Building on that new thinking, in 2011 Qualcomm went on to announce a program to release applications programming interfaces (APIs) to give web apps deeper links into its hardware. The APIs would be exposed directly into the browser, boosting performance and making the choice of OS less relevant, helping to drive cross-platform apps. AllJoyn is the most prominent platform tapping into this initiative as Qualcomm looks to build full developer environments around Snapdragon products and the burden it carries gets ever heavier. As CEO Paul Jacobs put it, ‘we’re trying to create the common language for the internet of everything’.
Power is important, but how will Qualcomm generate actual return on its hefty engineering investment in AllJoyn. Choudhury sees two important ways. One is actual Qualcomm products based on the technology, the first being AllPlay, a hardware/software media streaming product. For this firm, it is important to embed its software innovations in real chips and devices, because ‘we monetize through hardware’.
And at the recent Uplinq developer conference, AllJoyn also turned up in an experimental smart watch, called Toq, and in new chips from the Atheros division, providing a low power Wi-Fi platform for home appliances and consumer electronics, and eventually for IoT. The QCA4002 and QCA4004 integrate the IP stack and networking services on the chip to enable customers to add Wi-Fi to any product with minimal development effort or cost, said Qualcomm. An on-chip processor and memory avoid the need for a system controller, resulting in a package that is sufficiently simple and low cost to go into very basic connected gadgets. And customers can write their own applications on the platform, not just use it for connectivity.
The other return is more indirect, but comes from making the IoT simpler and more interoperability, and therefore encouraging it to expand rapidly ‘ creating the large scale which is necessary to make the economics of embedded gadgets and chips add up. Choudhury commented: ‘It’s not altruistic that we are doing this and giving it away, it’s to drive interoperability for this next wave of connected devices. Our business thrives on consumers, so we need scale. And this proximal interoperability is critical to growing businesses into new segments and markets. If consumers do not value connectivity, they will not buy these new devices, and that hurts our bottom line, because we monetize through processors, whether cellular or wireless and Bluetooth.’