© IGD 2010
Copyright iegd.institut.online.fr - 2010, tous droits réservés
3 questions to Philippe GAUTIER, by David Fayon
Philippe GAUTIER : There are three important ideas introduced at the same time, whose definitions are still subject to divergent opinions. Let's try to clarify this :
Historically, the Internet of Things -IOT- is a concept wherein objects are uniquely identified. These IDs are written on barcodes, RFID tags, NFC… then tagged on the physical objects. During the physical object lifecycle, event readings are made using sensors (RFID & NFC readers, cameras, scanners, GPS/GSM/Wi-Fi, manual reading, etc.). Those events are then collected and recorded in databases, and further aggregated, consolidated or merged with other information already handled in computer systems for traceability, logistics, management or sales issues.
As long as the processes are under control and well maintained, existing solutions work pretty well and provide satisfactory results. But, today, those approaches only fit the requirements of closed or semi-open loops or value chain management. Each implementation is segregated and cannot easily interoperate with others. Thus, maybe we should speak of a set of "Intranet of Things", since using the term Internet is improper.
However the primary ambition of the IoT was to publish these event related information on the Internet, in order to allow all actors involved with object manipulation or use to access them.
Various techniques and standards have appeared with the aim of publishing and retrieving this information, some of which are inherited from the existing Internet static resolution systems: ONS - Object Naming system - derived from DNS by the MIT and adopted by EPCGlobal. Other solutions allow for a more dynamic search, like what Google proposes on the Internet : a search engine that exhaustively indexes keywords found in natural language on the Web and manages to actually find (part of) the information looked for.
The Semantic Web technologies could, in theory, extend these search capabilities. Today, they allow to structure in a formal way the information published on the Web as metadata. This way of structuring can give contextual sense to published information and can be handled automatically in search engines (like in project “Wolfram Alpha”) or in exchanges between softwares.
Reusing and adapting current semantic web technologies in the Internet of Things therefore would allow anybody to automatically publish or search object-related information guaranteed to be directly usable and significant. I honestly don't know if this association exactly defines what will be the "Web 3.0", but I'm sure it can't easily succeed with the commonly used technologies and approaches.
Indeed, to my knowledge, all current approaches and proposed technologies rely on one or more pre established organization model: EPCGlobal, for instance, targets exhaustibility and universality for some businesses (CPG, automotive, aeronautics, pharmaceutical, etc.). On the other hand, the proposed metadata structure - for instance W3C’s Semantic Web one - are often little more than an arbitrary data organization.
However, whatever the exhaustibility of the model or its accuracy, it is always a subjective interpretation of things. This is even more so on the Web's complex ecosystem, where many different actors coexists - persons, companies, services, softwares… and now objects. What is good for some does not necessarily fit all.
I had, with my team, the opportunity to deploy IoT-related technologies, such as the EPCGlobal standards (ONS, EPCIS, etc.). We’ve seen those paradoxical situations where the same event-driven information could have different meanings. The information system could only see for what it was programmed, whilst each operational people could give it a distinct interpretation, according to the pursued goals at the time of the event. Human beings can make a context-rich judgment or can have a situated approach to the events but not existing software.
For example, package detection at the same RFID gantry could have had several meanings depending on whether a delivery was made, inventory reordering processed, theft carried out, reception triggered, unexpected run handled, etc.
In a controlled process, like in a company, it is possible to exhaustively program the information system to make it able to handle all different context situations and solve that interpretation issue of collected information. It is long and expensive, but feasible: specific implementations of ERPs are the “best” examples.
However, if you want to cover heterogeneous and complex organizations or target new uses such as it is on the continuously evolving Web, static organization models and metadata formal structures are unrealistic and futile, as there are too many autonomous, disparate and varied actors as well as different and heterogeneous situations. Therefore, other approaches are needed to undertake those goals. Only then it will be possible to consider and speak about IoT, semantic Web or even ... Web 3.0.
That way, sensor technologies will become these "invaluable tools" able to fully grasp the power of the information captured from the process reality. Today, they are only eyes, ears, noses or nerves that see, listen or touch, but without a brain able to smartly exploit information, evaluate and make appropriate decisions.
David Fayon: Regarding the Internet of Things, How to give objects some kind of software intelligence so that they can be able to act, react or operate according to the context? What could be the uses ?
Philippe GAUTIER : We’ve seen that, apart from controlled processes, any vertical or central approach (top down or functional) would be an endless and unsuccessful work. It doesn’t work in economy, politics, science, sociology… so why would it be different in the field of business process management, organization and information systems?
The only way to succeed is to take the problem from another angle and assess that sensor-generated information -retrieved from RFID, barcodes, GPS…- only makes sense in the context of the actor's end goals and purposes at a specific time, whoever is the actor: people, web services, objects, etc. We then must admit that objects can have goals, which is quite confusing today…but we’ll see this later.
Therefore, giving an event a meaning is, before anything, focusing on the context of pursued objectives at a given time: in the street, the information “a car comes from the left” means differently whether the user waits for his friend to pick him up or if he simply wants to cross the street. Consequently, further behavior of the user will be different.
So a Semantic web useful for the Internet of Things is not so much trying to create links between ideas or words but to be able to interpret the meaning of an information in a precise context : “what, where, when, how” and most importantly “why”. This requires a local autonomy in terms of perception, analysis, know-how and decision. In other words, this is no more than to put some intelligence at the lowest level (principle of subsidiarity).
On the existing Web, those clever actors are human beings and their social networks: this is the underlying concept of Web 2.0 as a distinction from Web 1.0.
Analogically, making objects true actors on the Web (3.0?) is no more than giving them enough autonomy and intelligence, suited to the roles we want them to play. Especially, we must give them the ability to move from one information ecosystem to another and accordingly pass from one information system to another seamlessly.
First obstacle: current hardware technology does not allow -for cost reasons- to embed such intelligence on an object. The price of an active RFID tag, associated with a memory and a processor, is disproportional, compared to the cost a pack of yogurt! On the other hand, Internet is ubiquitous - especially through mobile technologies - and enables us to deport such intelligence on different computers.
We then can now associate intelligent avatars - software objects - to each physical object, provided it meets either a need or a use and has an economic sense. Such avatars are virtual and independent software components, like “web services” and can be hosted on “cloud like” infrastructures, centralized servers, personal computers, smart phones, etc. They co-exist and co-evolve in symbiosis with their physical double: real objects. They support the uses associated with physical objects and have, consequently, specific aims. If properly designed, they can also move from one system to another corollary to the movement of material objects. Thus, they become their "virtual board computers”.
Accordingly, the objects, through their avatars, interact and interoperate together or with other actors of the Web: consumers, corporate computer systems, web services or sites, social networks ... according to the needs and uses. By gradually increasing the level of avatar’s intelligence, we then reach ubiquitous or pervasive computing, as Americans or Asian’s IOT practitioners often draw out.
Second obstacle: in order to do this, we need design methodologies and tools for software objects conception that are agnostical from both the further programming languages (Java / EE .NET, etc.), the patterns and naming standards (EPC, IPv6, URI, etc.) and the rigid and functional current approaches.
These methodologies and tools must also be used on an industrial scale: the Internet of Things’ market (or Web 3.0?) is potentially huge -the biggest ever known- but sometimes deals with quite simple things: consumer products, etc. It must then propose a standardized representation of both object structures (internal) and their behavior with each other (interaction) regardless of uses, trades, business models...
Finally, these design methodologies should be able to integrate and facilitate the generation of new services or practices: managing physical objects anywhere, anytime, anyhow; promote their reuse or sharing (vehicles, construction equipment and DIY, houses, caravans, etc.); facilitating the sharing of knowledge, backing the change management at any level, etc.
Is this “Alice in Wonderland” ? ... The challenges and stakes are, that with such avatars in place, the uses are endless:
E.G. : I, the suitcase in the airport, do I go to the right destination? Me, the book did I stay too long outside my library? I, the automatic trolley system, I’m asked, at the same time, to bring the X product in A Zone and the Y product in B Area. Given where I am, I start with the Y product. I, the airliner, I note that the boarded weight is greater than my builder’s tolerances. I refuse to take off and disobey to the pilot. I, the pack of yogurts, I know that my BBD will expire in 10 days. I then decide - without having been programmed for that - to discount me and, in coordination with the supermarket’s loyalty information system, notify known and present consumers on their mobiles phones by SMS.
But what matters in those examples is to clearly understand that these behaviors have not necessary been previously programmed by humans ... no one, either software designers or programmers, can predict all the scenarios or anticipate every opportunity in advance ... On the contrary, it is critical to understand that the avatar, by itself, can generate such behavior from its initial goals or autonomously generated new objectives: e.g. “providing security”, “be sold”, “ensuring its own integrity”, “not losing itself”, “monitoring its sharing”, etc.
David Fayon: Finally, could you introduce your company, Business2any, which edits a software engineering workshop for the Internet of Things - DOMIS, your methodology and your offer ?
Philippe GAUTIER : When I tell people that we have both a method and tools that allow those things, they first doubt ... I think the more skeptics are IT professionals, who have been taught that "programs only know or do what they have been programmed for”.
I am an IT professional since more than 20 years and I have never been satisfied with such a statement... Facing the stakes and promises of the Internet of Things and having to manage some related projects (EPCGlobal, RFID), I then had to search… and find alternate solutions. Having found one, particularly interesting, I then decided to create a company which delivers integrators, IT companies, IT departments, IOT pure players, etc. both a methodology and the associated tools to design and conceive IOT intelligent applications. I must confess that I had to find these solutions in very different areas from the original market of the Internet of Things, notably the military one…
We, at Business2Any, started from the fact that existing methods and tools for software design - too much inspired from purely functional approaches - hardly deal with complex situations or systems. And the Internet of Things (or Web 3.0) will be just billions of those various and unexpected uses or processes. The perfect example for functional description is UML. Anyone who has tried to model complex multi-agent systems using UML will understand my point…
To cope with, the approach is not to predict everything in advance then to model or describe exhaustively what should be reality, but to find ways to manage it as it will be: the key word is "monitoring"... In addition, the created applications must be scalable, that is to say, be able to climb up and to address new uses, while allowing those uses to emerge and auto catalyze. Finally, the applications must be able to process the information in specific contexts (semantic approach), to decide and act appropriately, including subsidiary levels in the organization. However, one important thing is to keep maintaining the convergence of goals if we want the system modeled at the global level remains consistent and operational.
To address those issues, Business2Any proposes a formal method, B-ADSC created by Dr. Janusz Bucki, a Franco-Polish Mathematician (expert in cybernetics) and a software engineering workshop - DOMIS - to help design, implementation and maintenance of software applications in the field of the Internet of Things (or Web 3.0).
DOMIS therefore helps to create those “avatars”, intelligent & autonomous software objects - or cyberobjets - that interact and collaborate, in parallel of their associated physical objects, on the Internet of Things. DOMIS also helps to jointly develop open organizations which host those avatars. Last, but not least, DOMIS can help working on the consistency and efficiency of the model and provides the software programmers with detailed and ready to use software specifications. We propose our solution: method and tools, to developers of the Internet of Things: IT consultants or departments, editors and integrators ... whatever their market segments: energy and sustainable development, home automation, logistics, transport pooling (car, bikes, trucks, all shared vehicles ...), 2/3G mobile (geo-positioning, real-time services), retail (merchandising, marketing), small services, industry, etc.
DOMIS, then, offers a standard for designing intelligent objects (internal structure, behavior, exchanges and interconnections) that allows physical objects:
To self-control: I, the pallet, am I still “sellable” regarding my own content and package? I, the shared bicycle, since I am standing there over two days, should I be checked ? ...
To implement their own policies: I, the pallet, I have stayed half an hour longer than expected in the shipping area. Can I tolerate it or not? Am I sensitive to temperature variations greater or smaller than 10 ° C ? ...
To communicate with their ecosystem: I, the pallet, I have stayed half an hour longer than expected in the shipping area, so I ask to come back into the storage area. The cold chain is broken, so I decide to switch my status to “unsellable” and notice to the logistic team...
Being "aware" of their preferences: I, the pallet, I'm “solicited” by two deliveries. Given the age of my content (BBD), I do prefer to include the delivery that matches the closer destination. I, the Smartphone of the car owner, I am solicited by three carpool while getting to my destination. Given their respective profiles, I choose one that matches the tastes of my owner...
Being "aware" of their “raison d'être”: I, the pallet, I’ve been moved 10 times in a day. So, without having been programmed for this, I give myself the goal of being controlled (support, hood) and alert the relevant operational people. I, patio table, I am aware of the bad weather to come. So, I give myself the goal of being put away for my own preservation. I, the new tire, I’ve been put on an axle where there is already a used tire. I reject this dangerous combination for the safety of the vehicle and warn the vehicle owner by SMS...
Finally, and crucially, DOMIS reduces software life cycle (design, development, testing, developments and dependability) and create, at an industrial scale, applications which can learn from feedback (artificial intelligence).
To conclude, I would say that the Internet of Things, the Semantic Web, or even a Web 3.0 will become a reality only if we change the way we design our applications. To do so, we must use formal methods that helps to treat issues and opportunities which will occur soon: the emergence of new organizations (business, communities), new capacity for self-organizing on distributed objects of everyday life, interactions between those cyberobjets, management and integration of feedback at the individual level (object) or group (business, communities), emergence of new value systems (economic, citizens ...), etc.. This evolution of the current Web will be processed in stages, hence the importance of having a formal methodology and tools of implementation that help to pass these levels. We deliver, from our side, solutions that help not "groping" and accomplish this double transformation.
Philippe Gautier, Head of Business2any, was previously Chief Information Officer. During this experience he was the first to implement all standards-based technologies "EPCGlobal" (GEN2 UHF RFID, EPCIS and ONS + DS developed within the EU funded BRIDGE project) in an innovative pilot and, further, at an operational level to manage - in a semi open-loop - the traceability of pallets amongst various logistics players in the supply chain. He earned the GS1 2005 Innovation price, the “Monde Informatique” SME 2006 Trophy and two "CIO 2007" trophies of “01 Informatique”: SMEs AND WINNER SPECIAL JURY PRIZE. He is a specialist of the B-ADSC method, the Internet of Things and author of numerous articles and chronicles. He speaks regularly in conferences and roundtables.
David Fayon : We can define Web 3.0 as the conjunction of the Internet of Things with the Semantic Web. What are the critical tools in this coming eco-system, sensor technologies such as RFID, barcodes or GPS on which the applications will be built ?