Distributed Statistical Analysis of Complex Systems Modeled Through a Chemical Metaphor Danilo Pianini ALMA MATER STUDIORUM Universit` a di Bologna, Italy danilo.pianini@unibo.it Stefano Sebastio IMT Institute for Advanced Studies Lucca Lucca, Italy stefano.sebastio@imtlucca.it Andrea Vandin Electronic and Computer Science University of Southampton, UK a.vandin@soton.ac.uk Abstract—The chemical-inspired programming approach is an emerging paradigm for defining the behavior of densely distributed and context-aware devices (e.g., in ecosystems of displays tailored to crowd steering, or to obtain profile-based coordinated visualization). Typically, the evolution of such sys- tems cannot be easily predicted, thus making of paramount importance the availability of techniques and tools supporting prior-to-deployment analysis. Exact analysis techniques do not scale well when the complexity of systems grows: as a conse- quence, approximated techniques based on simulation assumed a relevant role. This work presents a new simulation-based distributed analysis tool addressing the statistical analysis of such a kind of systems. The tool has been obtained by chaining two existing tools: MULTI VESTA and ALCHEMIST. The former is a recently proposed lightweight tool which allows to enrich existing discrete event simulators with automated and distributed statistical analysis capabilities, while the latter is an efficient simulator for chemical-inspired computational systems. The tool is validated against a crowd steering scenario, and insights on the performance are provided by discussing how the analysis tasks scale on a multi-core architecture. Keywords—Discrete Event Modeling and Simulation; Analysis of Emergent Behaviors; Automated Distributed Statistical Analy- sis; Statistical Model Checking; Chemical-Inspired Computation I. I NTRODUCTION The number of intercommunicating devices spread around the world is constantly increasing: sensors, phones, tablets, eyeglasses and many other everyday objects are carrying more and more computational and communicational capabilities. Such a computationally dense environment called for new programming approaches, many of them inspired by natural systems, e.g., biological [1], [2], physical [3] and chemical [4], [5]. In all of them, the overall system’s behaviour emerges from local, simple and probabilistic interactions among the devices composing the computational continuum. For this reason, most of the work in literature focuses on modelling single devices. When this reductionist point of view is adopted, a difficult task in the development methodology is to assert system properties: the system’s evolution can not be easily predicted and thus multiple simulation runs are performed [6]–[9]. Work supported by the European projects FP7-FET 256873 SAPERE, FP7- FET 257414 ASCENS, and FP7-STReP 600708 QUANTICOL, and by the Italian PRIN 2010LHT4KM CINA. Obvious questions accompany these procedures: how reliable are the obtained values? How is the number of performed simulations chosen? And how many simulations are required in order to state system properties with a certain degree of confidence? Moreover, there is frequently a lack of decoupling between the model specification and the definition of the system’s properties of interest: they are often embedded in the model, and their values are obtained via logging operations performed during the simulation process. This work presents a new tool obtained chaining ALCHEMIST [10], an efficient state-of-the-art simulator for chemical-inspired computational systems, with MULTI - VESTA [11], a recently proposed lightweight tool which allows to enrich existing discrete event simulators with automated and distributed statistical analysis capabilities. The result is thus a statistical analysis tool tailored to chemical-inspired pervasive systems. The benefits obtained by chaining the simulator with MULTI VESTA are: (1) a language (MULTI QUATEX) to compactly and cleanly express systems properties, decoupled from the model specification; (2) the automated estimation of the expected values of MULTI QUATEX expressions with respect to n independent simulations, with n large enough to respect a user-specified confidence interval; (3) an interactive plot as well as the generation of gnuplot input files to visualize the obtained results; (4) a client-server architecture to distribute simulations. The tool is validated by analyzing a crowd steering model reminiscent of the one presented in [12]. Synopsis. §II introduces ALCHEMIST and describes the crowd steering scenario, while §III outlines the main features of MULTI VESTA. Then §IV discusses the integration of the two tools, while §V validates the obtained tool. Finally, §VI reports some concluding remarks and future works. II. ALCHEMIST ALCHEMIST is a simulator targeting primarily chemical- inspired computational systems, namely distributed software systems that relies upon a chemical metaphor in order to obtain the desired global behavior. Various frameworks exist that are based on such metaphor, e.g., the Biochemical Tuple Spaces [2], Molecules of Knowledge [5] and SAPERE [4]. Some of the applications of such systems include display ecosystems 978-1-4799-5313-4/14/$31.00 ©2014 IEEE 416