Using ANFIS and FML for deploying Transparent Services in Smart Environments Giovanni Acampora, Vincenzo Loia and Autilia Vitiello Department of Computer Science University of Salerno 84084, Fisciano, Salerno, Italy Email: {gacampora, loia, avitiello}@unisa.it Abstract—This paper introduces a proposal for a smart environment capable of deploying enhanced services by using a synergic approach based on Adaptive Neuro-Fuzzy Inference System (ANFIS) and Fuzzy Markup Language (FML). In detail, this approach realizes a long-life learning strategy able to generate context-aware services and actualize them through abstraction techniques in order to maximize the users’ comfort and hardware interoperability level. As will be shown in experimental results, where a usability study has been performed, the proposed approach is capable of anticipating user’s requirements by automatically generating the most suit- able collection of interoperable services for improving users’ comfort. Keywords-Smart Environments; Ambient Intelligence; Fuzzy Systems; Neural Computing; Fuzzy Markup Language; I. I NTRODUCTION Smart environments are small worlds characterized by different kinds of intelligent and pro-active devices contin- uously interacting to make inhabitants’ lives more comfort- able and safe. In particular, these environments are aware of the specific features of human presence and behavior, take care of humans’ desires and are capable of responding intel- ligently to users’ initiative. In this new computing scenario, people should be able to seamlessly and unobtrusively use and configure devices composing their intelligent environ- ments without being cognitively and physically overloaded. Currently, smart environments are classified in three dif- ferent typologies: ∙ Virtual computing environments: they allow intelligent devices to access adequate services anywhere and any- time; ∙ Physical environments: they are enclosed by a variety of intelligent devices of different kinds such as tags, sensors and controllers and have sizes ranging from nano to micro to macro; ∙ Humans environments: humans, either individually or collectively, inherently form a smart environment for devices. However, humans may themselves be sup- ported by intelligent devices such as mobile phones and use surface-mounted devices (wearable computing) or contain embedded devices, e.g. pacemakers, operating in highly mobile nomadic environment [1]. In this paper, a novel methodology for designing generic smart environments is introduced. This approach realizes a long-life learning strategy able to generate context-aware services and actualize them through abstraction techniques in order to maximize the users’ comfort and hardware interoperability level. In order to achieve this objective, it has introduced an AmI Neuro-Fuzzy Computing System that exploits multi-agents paradigm, neural computing and fuzzy logic theory for designing a learning algorithm capable of capturing environmental features and users’ preferences in order to generate intelligent services that control the envi- ronment (in terms of lighting, temperature, blind opening, etc.) and satisfy users’ requirements. Our proposal defines the environmental intelligence concept through the so-called neuro-fuzzy services, i.e., adaptive services that exploits neural networks and fuzzy logic (ANFIS) to define high- quality relationships between services’ inputs and outputs in order to maximize users comfort. Moreover, in a such distributed framework, it appears clear that only those who succeeded in achieving hardware interoperability will be able to furnish innovative solutions and satisfy users’ needs. Our system exploits the FML technology to implement the inferred neuro-fuzzy services in an abstract way and, as a consequence, define a collection of so-called transparent neuro-fuzzy services. Experimental results section is devoted to validate our AmI approach through a usability study. In particular, we show how our proposal is capable of automatically generate a collection of fuzzy services which minimizes the gap between users requirements and systems actions, and, as consequence, leads to maximization the usability of the proposed framework. II. AMINEURO-FUZZY COMPUTING The AmI Neuro-Fuzzy Computing exploits multi-agents paradigm, neural computing and fuzzy logic in order to design an adaptive environment capable of capturing en- vironmental features mixed with users’ preferences and generate smart and interoperable intelligent services that satisfy human requirements. This result is achieved by defining a double-layered software architecture composed 2012 Sixth International Conference on Innovative Mobile and Internet Services in Ubiquitous Computing 978-0-7695-4684-1/12 $26.00 © 2012 IEEE DOI 10.1109/IMIS.2012.50 628