How Open Source initiatives can influence the Internet of Things

Alex JonssonBlogs

Tweet about this on TwitterShare on FacebookShare on LinkedInShare on Google+

open-source-hotel-palo-altoDuring the colorful IoT World event at the Palo Alto Crown Plaza earlier this Summer, a whole day was themed on open source for Internet of Things, moderated by industry expert Ajit Jaokar of Future Text. I was fortunate to attend the event and joined the action on for one of the panels on services and apps for IoT. I wanted to share a few thoughts and ideas on the current status of the open end of the industry, and some examples thereof.

I was a bit sceptical on the climate for using open source components for services and applications serving the industrial internet, as I touched down on the tarmac at SFO International airport before the event. With my experience and background in mobile, it is rare to hear otherwise fierce competitors speak freely on openness, collaborations on standards and work for the common good on least-denominator-type agreements at meetups, interleaved with ranting debates about which operating system or software stack is more open than the next. There are conferences with claims to address open technologies, processes, formats and common albeit lame API definition proposals; but in the end, it’s mostly about men in suits holding cordial open discussions about closed boxes with thin API layers at best – claiming that any major part of of any currently available mobile ecosystem carries a healthy degree of openness, is mostly about marketing positioning.

With my historical handicap in the mobile scene, I was happily surprised in meeting many fine industry experts, researchers and engineers who spend their days on making things better connected and less closed. The now classic Beecham Research primer illustration of IoT, the explode pie chart with its multitude of colorful slices, has proven to be realistic enough to describe where the industry has been at for some time. Each pie slice has its formats, standards, three-letter abbreviations and corresponding set of challenges for creating services; there are also common grounds between sectors, such as the use of sensors, filtering at various levels, aggregation, & presentation of data and various levels of interaction and  closing process loops of what the users need and want.

iot segment

One of these slices, not so industrial as many others,  is related to the quantified self, health monitoring and fitness tracking mostly for the consumer market. It provides a good example of the type of initiatives are at currently at work. On one end there are researchers on the bleeding edge like Naveen Selvadurai, who has opted-in to liberate his own personal data, thereby creating a Human API of sorts; while commercial players such as Fitbit, Jawbone and even more so Nike/Apple has more the opposite in mind – with various degrees of silo mindsets with the distant goal of owning the user’s wrist. To host open APIs for services offerings, commercial or not, as opposed to liberating and aggregating data from silos, you don’t need to get all the fitness trackers around and wearing them concurrently as a method of ensuring data ownership, as has been mentioned in the Ars Technica’s article by Casey Johnston.

At the conference, other efforts involving aggregating sensor data from humans included open source projects such as The Locker Project, offering an open virtual data locker for storing human sensor data coupled with the visualization tools offered by Fluxtreme. There were also more traditional service offerings mentioned, like Xively (formally Pachube, Cosm) and Runkeeper where while the data formats themselves are open the service and data storage of such data is closed. Last year when online data storage Zeo closed its last data port, all the data aggregated by active people all around the world needed to be exported or sent to oblivion. As a general remark, owning your own data, whether you’re a citizen, a company, a municipality, region or country and safely controlling access – was one of the major themes throughout the conference, mentioned by representatives of many ecosystem roles.

Another important theme is the establishment of standards for information interchange between things, aggregation points, services and the end user; the industry initiatives Alljoyn (of the allseenalliance.org, initiated by Qualcomm) and Eclipse IoT show good speed forward, for the discovery and interoperability across the entire IoT sector. The efforts origins started from brushed-up, legacy telemetry standards like MQTT (IBM/Eurotech, invented in 1999), or technology transfers from other industries – or in our analogy other pie sectors such as XMPP (Extensible Messaging and Presence Protocol, formally Jabber) which originally started as a chat service that ran over TCP connections is now offered over HTTP in various implementations where traffic can be routed via standard web servers. I intend to cover the specific innards of the various formats in a sequel post and what we’re actually looking for in formats for data transport, storage and dissemination to ensure fragmentation doesn’t become a major obstacle for IoT application developers.

To sum it up, there is a healthy number of open source initiatives; several of which also had representation live at the Crown Plaza. Whereas the optimum between security, ease of use and interoperability is yet to be found – open source technologies are positioned well to make up some of the building blocks. Already, most links in the value chain is represented from near-sensor, including open hardware architectures to client applications running IoT services over phones and tablets are well defined. There is good reason to keep a watchful eye on what open technologies can contribute with in this heterogeneous industry of many origins, not having to reinvent the wheel for each slice of the IoT industry every time.

Below a list, as compiled by industry guru Ajit Jaokar, for some of the noteworthy initiatives to date: