Page 33 - SMILESENG

Intl. Summer School on Search- and Machine Learning-based Software Engineering
Discovery Service Federation in the Web of Things
Juan Alberto Llopis Applied Computing Group (TIC-211), University of Almer´ıa Almer´ıa, Spain
jalbertollopis@ual.es
Javier Criado
Applied Computing Group (TIC-211), University of Almer´ıa Almer´ıa, Spain
javi.criado@ual.es
Luis Iribarne
Applied Computing Group (TIC-211), University of Almer´ıa Almer´ıa, Spain
luis.iribarne@ual.es
Abstract—As the number of technological devices in our environment increases, the need for mechanisms to facilitate their use becomes more evident. A large part of these devices are Internet of Things (IoT) devices, devices that are accessible through a wide set of technologies, protocols and applications. To homogenize the representation, communication and management of IoT systems, web technology offers a series of mechanisms that facilitates the construction of heterogeneous systems. Inspired by this idea, the Web of Things (WoT) concept is an IoT approach that uses web technology to create a representation of physical entities in digital space. To perform the integration between both spaces, it is necessary to identify the devices or physical entities deployed and to establish a communication between the physical and digital space. Therefore, there is a need for a mechanism that allows applications and entities to obtain or discover services that meet their requirements. In this PhD thesis, we propose a federated discovery service for the WoT, capable of finding and registering WoT devices within reach of the discovery service or available in other discovery services. In a federated discovery service, when searching for devices, a device can be located in the same discovery service or other discovery services. Therefore, a criterion has to be established to select between (a) devices in a discovery service, (b) devices in different discovery services; and (c) discovery services. The proposed discovery service uses a recommender system to suggest devices that best suit the user’s requirements. This work can reduce the workload of the development of Smart solutions by facilitating the discovery of WoT devices and by reducing the use of energy due to device response calls.
I. INTRODUCTION
With the increase in the number of deployed Internet of Things (IoT) devices and with the integration of the physical and digital space to represent physical entities on the web, searching for devices has become more difficult [1]. The Web of Things (WoT) is an approach that homogenizes the representation, communication and management of IoT sys- tems, facilitating the interaction between devices and external entities [2]. However, there is still a problem when looking for devices, due to the increasing number of devices, the use of techniques to improve the search process is desired.
When searching for services, discovery services were de- veloped to facilitate the search for services that meet certain requirements among a large number of services offered [3].
This work has been funded by the EU ERDF and the Andalusian Govern- ment (Spain) under the project I+D UrbanITA, ref. PY20 00809. Juan Alberto Llopis has been funded by a grant of the Spanish Government FPU19/0072.
Discovery services make use of a repository or directory that stores the descriptions of the set of services it manages, avoid- ing the need to define them multiple times, and facilitating their access and use by interested applications or entities [4]. This reduces the number of requests made directly on devices, reducing the energy usage of the devices and increasing their usage time [5]. Discovery services in the field of service- oriented architectures (SOA) are a mechanism widely used by the community [6], [7]. In the IoT domain, there are also approaches that provide some middleware and Content Distribution Networks (CDNs) for service discovery [8], [9]. However, there are still no discovery services in the WoT domain that take into account needed features in these systems such as interoperability, security or service heterogeneity, among other possible examples [10], [11].
In this PhD thesis, we propose a discovery service federation for the WoT, capable of delegating and extending queries to other discovery services when a discovery service is not able to answer a query. In this way, the proposed discovery service will not only allow identification, classification and use of the available entities but will also allow the discovery of these entities through other discovery services. In addition, a discovery service can be ineffective if the queries are not constructed correctly. Therefore, the research analyzes techniques to perform query mutation, so that a query that does not obtain results can be automatically modified by creating variants of that query to discover devices.
Finally, based on a request sent by a client, and taking into account the matching conditions, a discovery service returns a set of devices that meet the client’s requirements. The returned set of devices can be ordered, prioritized, and/or filtered to allow the selection of the best alternative. Such selection may be based on criteria that look at various features, such as user preferences, usage history, or other information of interest to the application domain. For this reason, this PhD thesis will incorporate the application of recommender systems to be able to suggest the best alternatives during service discovery [12]. Figure 1 shows the proposed linkage between federated discovery systems and recommender systems. The recommender system will use Artificial Intelligence (AI) to recommend devices from queries in the form of a natural language sentence.
21