Challenge: Distributed Sensing Applications in HighlyHeterogeneous and Multimodal Pervasive EnvironmentsMario Di FrancescoWhitepaperNSF Workshop on Pervasive Computing at Scale (PeCS)Background Dr. Di Francesco has been working in the field of Wireless Sensor Networks(WSNs) for several years, with specific focus on energy-efficient mechanisms for data collec-tion and dissemination [1]. He investigated how different sensing modalities and acquisitionstrategies can improve the quality of information and the energy-efficiency in WSNs [2]. He alsoconsidered urban sensing scenarios and, more generally, applications exploiting mobile elementsin WSNs [3, 4]. In this context, he applied learning algorithms for improving the efficiency ofdata collection [5]. More recently, he started investigating privacy issues in pervasive systems,with special reference to smart environments [6].Vision Pervasive applications have become increasingly complex. Context and situation aware-ness – complemented by learning and data mining algorithms – have emerged as effective ab-stractions to map data (which are often noisy or too abundant) to meaningful representation ofstates, activities, and patterns. Rich characterization of contextual information can significantlybenefit from scale, which brings fine resolution and extended coverage. Since it is impractical toexplicitly instrument large-scale pervasive environments, an intriguing option consists in exploitingwhatever devices – sensing platforms, application-specific devices already present in the environ-ment (such as surveillance cameras), personal communication devices (such as smartphones), andso on – are incidentally available in the environment to perform distributed applications. Thesedevices should self-organize to collaboratively allocate and execute even multiple applications atthe same time. Since nodes spontaneously interact, there is no control not only on the networkarchitecture, but also on the specific kind (or family) of the participating devices, which are po-tentially highly heterogeneous. However, heterogeneity should not be seen as a foe, but rather asa resource to federate specialized (hence efficient) devices in a pervasive infrastructure which canexploit rich interactions by means of multimodal sensing and multi-paradigm communication.Significance While many approaches to pervasive applications operates on top of a specificnetwork architecture or communication paradigm, much effort has been also targeted to bridgediverse (and sometimes technically incompatible) technologies. A recent research direction relatesto the field of the so-called Internet-of-Things [7]. The heterogeneity of devices is addressed inthis context, however the focus is still on interconnection issues, in terms of both homogeneityof in the internetworking and the availability of standardized access to data [8]. Hence, anapproach targeted at giving value to heterogeneity (in terms of functions), while at the sametime abstracting from the differences between the individual devices, is definitively a foundation oflarge-scale pervasive environments. In fact, it will ease the creation and deployment of distributedapplications, as well as improving the efficiency and the robustness of the system.Handling heterogeneity requires to rethink conventional paradigms, for instance those used inthe context of middleware architectures for sensing devices [9]. In fact, the following significantchallenges arise.Heterogeneous devices imply a variety of platforms, architectures, and operating systems.A framework for heterogeneous devices should be carefully designed to meet the contrastingrequirements of abstracting the functions of the nodes and providing access to the featurespeculiar to individual devices.How to obtain a distributed application suitable to pervasive environments is demanding,since the decomposition in tasks has to be aware of the heterogeneity of devices. Hence,new methodologies and tools have to be developed, with particular focus on optimizing theresource utilization.The large number of devices leads to scalability and reliability issues. Also here heterogeneitycan help: devices can use different communication technologies, and in some cases evenmultiple interfaces at the same time, thus increasing the overall spectrum utilization andraw bandwidth. However, proper scheduling and routing mechanisms have to be defined.Due to the large scale of the pervasive environment, security threats are critical, sincethey might spread over the entire system very rapidly, and with very serious consequences.Hence, attacks or selfish/malicious behaviors should be prevented and controlled as muchas possible from the early stages of their development.References[1] G. Anastasi, M. Conti, M. Di Francesco, and A. Passarella. Energy conservation in wireless sensornetworks: A survey. Ad Hoc Networks, 7(3):537–568, May 2009.[2] C. Alippi, G. Anastasi, M. Di Francesco, and M. Roveri. A survey on energy management in wirelesssensor networks with energy-hungry sensors. IEEE Instrumentation and Measurements Magazine,12(2):16–23, April 2009.[3] G. Anastasi, M. Conti, and M. Di Francesco. Reliable and energy-efficient data collection in sparsesensor networks with mobile elements. Performance Evaluation, 66(12):791–810, December 2009.[4] M. Di Francesco, S. K. Das, and G. Anastasi. Data collection in wireless sensor networks withmobile elements: A survey. ACM Transactions on Sensor Networks, to appear.[5] M. Di Francesco, K. Shah, M. Kumar, and G. Anastasi. An adaptive strategy for energy-efficientdata collection in sparse wireless sensor networks. In Proceedings of the 7thEuropean Conferenceon Wireless Sensor Networks (EWSN 2010), pages 322–337, February 2010.[6] G. Pallapa, M. Di Francesco, and S. K. Das. Adaptive and context-aware privacy preservationschemes exploiting user interactions in smart environments. Submitted to the 9thIEEE InternationalConference on Pervasive Computing and Communications (PerCom 2011).[7] Hakima Chaouchi, editor. The Internet of Things: Connecting Objects. Wiley-ISTE, 2010.[8] A.P. Castellani, N. Bui, P. Casari, M. Rossi, Z. Shelby, and M. Zorzi. Architecture and protocolsfor the internet of things: A case study. In Pervasive Computing and Communications Workshops(PERCOM Workshops), 2010 8th IEEE International Conference on, pages 678–683, april 2010.[9] Karen Henricksen and Ricky Robinson. A survey of middleware for sensor