In the last years, researchers are devoting many efforts to improve technological aspects of the Internet of Things (IoT), while little attention has dedicated to social and practical sides. Professional developers program the behavior of smart objects. In addition, often the functionality exposed by a single object are not able, alone, to exhaustively support the end users' tasks. The opportunities offered by IoT can be amplified if new highlevel abstractions and interaction paradigms enable also non-technical users to compose the behavior of multiple objects. To fulfill this goal, we present a model to express rules for smart object composition, which includes new operators for defining rules coupling multiple events and conditions exposed by smart objects, and for defining temporal and spatial constraints on rule activation. Such model has been implemented in a Web application whose composition paradigm has been designed during an elicitation study with 25 participants.
The current integrated technologies confer intelligence to any type of objects (also called things in this document) and connect them to the network. For example, a pressure sensor can make an office door handle smart, for example alerting a user located in another part of the world every time someone comes into that office. The added value is that the object, being part of a network of smart things (the Internet of Things (IoT) can communicate with other objects or services, thus triggering dynamic behavior. If the door sensor is able to communicate with a sensor worn by a person, than a remote user can also know who entered the office.
possibility to collect in real time information concerning events and behaviors happening in the real world. For example, a potential advantage of the IoT is to anticipate the needs of a human even before he is aware of it. Today these scenarios may appear unrealistic since we are not pervaded with a significant number of sensors. However, Cisco’s forecasting estimates that more than 50 billion smart objects will be deployed by 20201. There are already many contexts in which the IoT is adopted. For example, for creating wearable devices, i.e. clothing and accessories such as bracelets, watches, t-shirts, rings, shoes, can integrate sensors capable of detecting physiological parameters and actuators to communicate with the users. Domotics is another very active IoT context: pressure, volumetric and distance sensors, as well as infrared cameras, are able to ensure the security of homes; they permit a remote user to interact with the devices installed in the house (e.g., lights, heating, blinds) and assist user to avoid hazards such as floods, fires or explosions. Other sectors that are beneficiating of the advent of IoT are Smart Cities, industry and environment through energy saving.
content?type=webcontent&articleId=1208342
From the technological point of view, such a large number of things requires an adequate network infrastructure and efficient communication protocols, especially due to the fact that integrated devices have very limited resources (e.g., CPU, RAM, memory, battery). Further issues, such as privacy (e.g., in the smart door scenario, if and when it is allowed to know who entered into the office) and safety (e.g., if a hacker is able to break a device, then it could access all the devices it is connected with) must be addressed.
From the HCI point of view, a major challenge is to enable
even non-technical users to manipulate data and
functionality of things in a simple and natural way. Today, in
fact, this is a prerogative almost always reserved to
developers who, through the use of specific programming
languages, provide pre-packaged solutions to users. The
most important challenge is to allow non-technical users to
define and manage the connections between things, which
represent the real benefit of IoT, especially in the next years
when a growing amount of things will be available.
Some works in the literature propose mashup techniques for
addressing this issue. For example in [
As mentioned above, things can be treated as services,
because sensors and actuators have an URI that identify
them on the Internet. In the case of service mashups,
platforms implementing an event-driven approach, as the
one described in [
In fact, most of these platforms currently used in the IoT domain do not permit (for example IFTTT) or make it difficult (for example gluethings) to specify multiple events and actions. Similar difficulties arise when specifying temporal or spatial constraints, e.g., to define the following behavior: “If I’m in Rome and I post an image on Instagram between 8.00 and 11.00 a.m., post the same picture also on Twitter and Tumblr”.
In order to identify composition paradigms able to guide
users in the definition of articulated rules, we wanted to
elicit the end-user mental model, which is an aspect scarcely
explored in the field of task automation, as pointed out by
[
In this section we illustrate through an example the main features of the 5Ws composition paradigm. A user, who we suppose is a female, creates a rule to automatically turn on the coffee machine and roll-up the shutters when her smart bracelet detects that she has just woken up or the smart alarm clock rings. To create this rule, the user clicks the “New Rule” button in the navigation bar (Errore. L'origine riferimento non è stata trovata., circle 1) and the “Creating Rule” interface appears. The UI shows the main area in which a rule is defined. The left side is for specifying the triggering events, and the right side is to define the actions to be activated by the selected services.
A wizard procedure, activated by the green “+” button highlighted by circle 2 in Errore. L'origine riferimento non è stata trovata., guides the users in defining the events in a full-automated fashion. The wizard sequentially shows some pop-up windows in which the service, the events and the conditions are specified. According to WYSIWYG approach, the wizard steps allow the user to define an event in terms of Which is the service to be monitored for detecting the triggering event, What service event has to be monitored, When and Where the event has to occur. The specification of When and Where conditions is optional. At the end of the wizard procedure, the event is defined and its summary appears under the “Events” area. Actions can be defined by clicking on the green “+” button highlighted (circle 3 in Errore. L'origine riferimento non è stata trovata.). The button activates a wizard that helps the user define an action in terms of Which service will execute the action as a consequence of the event(s), What action the service has to perform and When and Where the action can be performed.
In EFESTO, users may either define first all the events and then the actions, or define first a basic rule with one event and one action and then include new events and new actions. Events and actions can be added or removed at any time fostering a dynamic modification of the running mashup. Further events can be added by clicking one of the two green “+” buttons labeled And / Or. Choosing the “And” button starts the definition of a new event that will cause the execution of the rule action(s) if all conditions of all events are satisfied. The “Or” button determines the definition of a new event that will cause the execution of the rule action(s), if the conditions of at least one event are satisfied. Once the rule is created (see Figure 2), it can be saved by entering a short description of the rule (the Why in the 5W model).
One of the cornerstones of the future of the IoT will be to put in the hands of the end users simple software tools capable of making natural and powerful composition between things. It is unimaginable that this possibility is reserved for a few experts. As already happened in the past with software tools (forums, social networks, CMS) that made users to evolve from simple consumer to prosumer, even for the IoT have to be designed tools suitable even for non-expert users. This paper has illustrated how a generic platform for service mashups can be specialized for the composition of services that enable accessing/controlling smart things. We are currently working on implementing the needed extensions. A demo, illustrating some preliminary results, will be given during the workshop.