In the realm of smart cities, telecommunication companies (telcos) are expected to play a protagonistic role as these can capture a variety of natural phenomena on an ongoing basis, e.g., traffic in a city, mobility patterns for emergency response or city planning. The key challenges for telcos in this era is to ingest in the most compact manner huge amounts of network logs, perform big data exploration and analytics on the generated data within a tolerable elapsed time. This paper introduces SPATE, an innovative telco big data exploration framework whose objectives are two-fold: (i) minimizing the storage space needed to incrementally retain data over time, and (ii) minimizing the response time for spatiotemporal data exploration queries over recent data. The storage layer of our framework uses lossless data compression to ingest recent streams of telco big data in the most compact manner retaining full resolution for data exploration tasks. The indexing layer of our system then takes care of the progressive loss of detail in information, coined decaying, as data ages with time. The exploration layer provides visual means to explore the generated spatio-temporal information space. We measure the efficiency of the proposed framework using a 5GB anonymized real telco network trace and a variety of telco-specific tasks, such as OLAP and OLTP querying, privacy-aware data sharing, multivariate statistics, clustering and regression. We show that out framework can achieve comparable response times to the state-of-the-art using an order of magnitude less storage space.

In this demonstration paper, we present SPATE, an innovative telco big data exploration framework whose objectives are two-fold: (i) minimizing the storage space needed to incrementally retain data over time, and (ii) minimizing the response time for spatiotemporal data exploration queries over stored data. Our framework deploys lossless data compression to ingest streams of telco big data in the most compact manner retaining full resolution for data exploration tasks. We augment our storage structures with decaying principles that lead to the progressive loss of detail as information gets older. Our framework also includes visual and declarative interfaces for a variety of telco-specific data exploration tasks. We demonstrate SPATE in two modes: (i) Visual Mode, where attendees will be able to interactively explore synthetic telco traces we will provide, and (ii) SQL Mode, where attendees can submit custom SQL queries based on a provided schema.

The demand for indoor localization services has led to the development of techniques that create a Fingerprint Map (FM) of sensor signals (e.g., magnetic, Wi-Fi, bluetooth) at designated positions in an indoor space and then use FM as a reference for subsequent localization tasks. With such an approach, it is crucial to assess the quality of the FM before deployment, in a manner disregarding data origin and at any location of interest, so as to provide deployment staff with the information on the quality of localization. Even though FM-based localization algorithms usually provide accuracy estimates during system operation (e.g., visualized as uncertainty circle or ellipse around the user location), they do not provide any information about the expected accuracy before the actual deployment of the localization service. In this paper, we develop a novel frame-work for quality assessment on arbitrary FMs coined ACCES. Our framework comprises a generic interpolation method using Gaussian Processes (GP), upon which a navigability score at any location is derived using the Cramer-Rao Lower Bound (CRLB). Our approach does not rely on the underlying physical model of the fingerprint data. Our extensive experimental study with magnetic FMs, comparing empirical localization accuracy against derived bounds, demonstrates that the navigability score closely matches the accuracy variations users experience.

In this demonstration we present ACCES, a novel framework that enables quality assessment of arbitrary fingerprint maps and offline accuracy estimation for the task of fingerprint-based indoor localization. Our framework considers collected fingerprints disregarding the physical origin of the data. First, it applies a widely used statistical instrument, namely Gaussian Process Regression (GPR), for interpolation of the fingerprints. Then, to estimate the best possibly achievable localization accuracy at any location, it utilizes the Cramer-Rao Lower Bound (CRLB) with interpolated data as an input. Our demonstration entails a standalone version of the popular and open-source Anyplace Internet-based indoor navigation service in which the software modules of ACCES are integrated. At the conference, we will present the utility of our method in two modes: (i) Collection Mode, where attendees will be able to use our service directly to collect signal measurements over the venue using an Android smartphone, and (ii) Reflection Mode, where attendees will be able to observe the collected measurements and the respective ACCES accuracy estimations in the form of an overlay heatmap.

Predominant smartphone OS localization subsystems currently rely on server-side localization processes, allowing the service provider to know the location of a user at all times. In this paper, we propose an innovative algorithm for protecting users from location tracking by the localization service, without hindering the provisioning of fine-grained location updates on a continuous basis. Our proposed Temporal Vector Map (TVM) algorithm, allows a user to accurately localize by exploiting a k-Anonymity Bloom (kAB) filter and a bestNeighbors generator of camouflaged localization requests, both of which are shown to be resilient to a variety of privacy attacks. We have evaluated our framework using a real prototype developed in Android and Hadoop HBase as well as realistic Wi-Fi traces scaling-up to several GBs. Our study reveals that TVM can offer fine-grained localization in approximately four orders of magnitude less energy and number of messages than competitive approaches.

A wide spectrum of Internet-scale mobile applications, ranging from social networking, gaming and entertainment to emergency response and crisis management, all require efficient and scalable All k Nearest Neighbor (AkNN) computations over millions of moving objects every few seconds to be operational. In this paper we present Spitfire, a distributed algorithm that provides a scalable and high-performance AkNN processing framework to our award-winning geo-social network named Rayzit. The proposed algorithm deploys a fast load-balanced partitioning along with an efficient replication-set selection, to provide fast main-memory computations of the exact AkNN results in a batch-oriented manner. We evaluate, both analytically and experimentally, how the pruning efficiency of the Spitfire algorithm plays a pivotal role in reducing communication and response time up to an order of magnitude, compared to three state-of-the-art distributed AkNN algorithms executed in distributed main-memory.

A wide spectrum of Internet-scale mobile applications, ranging from social networking, gaming and entertainment to emergency response and crisis management, all require efficient and scalable All k Nearest Neighbor (AkNN) computations over millions of moving objects every few seconds to be operational. Most traditional techniques for computing AkNN queries are centralized, lacking both scalability and efficiency. Only recently, distributed techniques for shared-nothing cloud infrastructures have been proposed to achieve scalability for large datasets. These batch-oriented algorithms are sub-optimal due to inefficient data space partitioning and data replication among processing units. In this paper, we present Spitfire, a distributed algorithm that provides a scalable and high-performance AkNN processing framework. Our proposed algorithm deploys a fast load-balanced partitioning scheme along with an efficient replication-set selection algorithm, to provide fast main-memory computations of the exact AkNN results in a batch-oriented manner. We evaluate, both analytically and experimentally, how the pruning efficiency of the Spitfire algorithm plays a pivotal role in reducing communication and response time up to an order of magnitude, compared to three other state-of-the-art distributed AkNN algorithms executed in distributed main-memory.

Wi-Fi (or WLAN) based indoor navigation applications for mobiles rely on cloud-based services (s) that take care of a user's (u) localization task using structures called Radio Maps (RMs). It is imperative for u to have a stable WiFi connection in order to either continuously receive location updates from s or to download RMs a priori for offline navigation. Wi-Fi networks however, suffer from intermittent connectivity due to poor network planning that results in sparse deployment of access points and effectively areas where Wi-Fi coverage cannot be guaranteed. This inherently affects the localization accuracy and therefore the navigation experience of users. In this paper, we propose an innovative framework for accurate and fast indoor localization over an intermittently connected WiFi network, coined Prefetching Localization (PreLoc). In Preloc, we prioritize the download of RM records based on knowledge acquired from historic traces of other users inside the same building. Instead of downloading the complete RM from s to u, we propose a Probabilistic Group Selection (PGS) strategy, which identifies RM records that have a higher probability of being necessary to a user moving inside a target area. We have evaluated our framework using a real prototype developed in Android, as well as realistic Wi-Fi traces we collected at the University of Cyprus. Our experimental study reveals that PreLoc using PGS and conventional fingerprint-based indoor positioning algorithms can yield accuracy that is as good as using the same algorithms with a complete RM, even under scenarios of weak Wi-Fi coverage.

Abstract—The smartphone revolution has introduced a new era of social networks where users communicate over anonymous messaging platforms to exchange opinions, ideas and even carry out commerce. These platforms enable individuals to establish social interactions between strangers based on a common interest or attribute. In this paper we present Rayzit1, a novel anonymous crowd messaging architecture, which utilizes the location of each user to connect them instantly to their k Nearest Neighbors (kNN) as they move in space. Contrary to the very large body of location-based social networks that suffer from bootstrapping issues, our architecture enables a user to always interact with the geographically closest possible users around. We establish this communication using a fast computation of an All kNN query that generates a dynamic global social graph every few seconds. We present motivating application scenarios and the detailed back-end architecture that allows Rayzit to scale. We have collected and analyzed data from the interactions of thousands of active users and confirm our claims. Keywords—Crowd, Social, Mobile, Microblogging I.

People do most of their activities, business, commerce, entertainment and socializing indoors. As all of these are increasingly aided by online services and indoor spaces are becoming bigger and more complex, there is a growing need for cost-effective indoor localization, mapping, navigation and information services. In this paper, we present a complete Indoor Information Service, coined Anyplace, which has an open, modular, extensible and scalable architecture, making it ideal for a wide range of applications. Our service features three highly desirable properties, namely crowd sourcing, scalability and accuracy. Anyplace implements a set of crowd sourcing-supportive mechanisms to handle the enormous amount of crowd-sensed data, filter incorrect user contributions and exploit Wi-Fi data from heterogeneous mobile devices. Moreover, it uses a big-data architecture for efficient storage and retrieval of localization and mapping data. Finally, our service relies on the abundance of sensory data on smartphones (e.g., Wi-Fi signal strength and inertial measurements) to deliver reliable indoor geolocation information that received several international awards.

Indoor Positioning Systems (IPS) have recently received considerable attention, mainly because GPS is unavailable in indoor spaces and consumes considerable energy. On the other hand, predominant Smartphone OS localization subsystems currently rely on server-side localization processes, allowing the service provider to know the location of a user at all times. In this paper, we propose an innovative algorithm for protecting users from location tracking by the localization service, without hindering the provisioning of fine-grained location updates on a continuous basis. Our proposed Temporal Vector Map (TVM) algorithm, allows a user to accurately localize by exploiting a k-Anonymity Bloom (kAB) filter and a bestNeighbors generator of camouflaged localization requests, both of which are shown to be resilient to a variety of privacy attacks. We have evaluated our framework using a real prototype developed in Android and Hadoop HBase as well as realistic Wi-Fi traces scaling-up to several GBs. Our analytical evaluation and experimental study reveal that TVM is not vulnerable to attacks that traditionally compromise k-anonymity protection and indicate that TVM can offer fine-grained localization in approximately four orders of magnitude less energy and number of messages than competitive approaches.

Crowdsourcing refers to a distributed problem-solving model in which a crowd of undefined size is engaged in the task of solving a complex problem through an open call. This novel problem-solving model found its way into numerous applications on the web for voting, fund-raising, micro-works and wisdom-of-the-crowd scenarios. On the other hand, the shift of desktop users to mobile platforms in the post-PC era, along with the unique multi-sensing capabilities of modern mobile devices are expected to eventually unfold the full potential of Crowdsourcing. Smartphones offer a great platform for extending and diversifying web-based crowdsourcing applications to a larger contributing crowd, making contribution easier and omni-present. This advanced seminar presents the fundamental concepts behind crowdsourcing and its applications to mobile data management. In the first part of the seminar, we will overview the crowdsourcing research landscape from a variety of perspectives, with a particular emphasis on the latest data management trends. In the second and more extended part of the seminar, we will focus on an in-depth coverage of emerging mobile crowdsourcing architectures and systems, through a multi-dimensional taxonomy that will address location, sensing, power, performance, big-data and privacy among others. Furthermore, we will overview a number of in-house crowdsourcing prototypes we have developed and deployed over the last few years. The seminar concludes with challenges, opportunities and new directions in the field.

In this paper we present our Crowdsourced Linked Open Data Architecture (CLODA), a first attempt to combine crowdsourcing, localization and location-based services to generate, collect, validate and relate real-world, geo-spatial and multidimensional information using smartphones and other mobile devices. CLODA focuses on the construction of URI addressable, interlinked and semi-structured data following the Linked-Open Data (LOD) paradigm. The validity of the constructed data is then contributed by a participating crowd. We present our prototype implementation on top of Google Maps and a blend of in-house technologies, particularly our indoor positioning framework, coined Airplace, our trajectory similarity framework, coined SmartTrace, our neighborhood detection framework, coined Proximity and our smartphone testing platform coined SmartLab.

Consider a centralized query operator that identifies to every smartphone user its k geographically nearest neighbors at all times, a query we coin Continuous All k-Nearest Neighbor (CAkNN). Such an operator could be utilized to enhance public emergency services, allowing users to send SOS beacons out to the closest rescuers, allowing gamers and social networking users to establish ad-hoc overlay communication infrastructures, in order to carry out complex interactions. In this paper, we study the problem of efficiently processing a CAkNN query in a cellular or WiFi network, both of which are ubiquitous. We introduce an algorithm, coined Proximity, which answers CAkNN queries in O(n(k + λ)) time, where n denotes the number of users and λ a network-specific parameter (λ <;<; n). Proximity does not require any additional infrastructure or specialized hardware and its efficiency is mainly attributed to a smart search space sharing technique we introduce. Its implementation is based on a novel data structure, coined k+-heap, which achieves constant O(1) look-up time and logarithmic O(log(k*λ)) insertion/update time. Proximity, being parameter-free, performs efficiently in the face of high mobility and skewed distribution of users (e.g., the service works equally well in downtown, suburban, or rural areas). We have evaluated Proximity using mobility traces from two sources and concluded that our approach performs at least one order of magnitude faster than adapted existing work.

The focus of this work is to adaptively optimize continuous queries in a wireless sensor network. Therefore, we analyze and categorize the different types of queries used by wireless sensor applications. We present a short survey on the existing algorithms that can be adapted and used to optimize the different types of queries. We present their strengths and their drawbacks and identify that continuous subset queries can be further optimized. We propose a framework for answering continuous subset queries with multiple operators that can adapt to network and data changes. A query routing tree can be broken down into an operator tree, an operator placement assignment and a routing scheme. The operator tree defines the order in which the operators will be applied to the data and how the results will be pipelined from one operator to the other. The operator placement assignment defines on which node of the network each operator will be hosted and executed. We assume shortest-path routing and propose efficient methods to adapt operator placement and the operator tree dynamically for continuous queries. Our technique for adapting the actual operator tree is the first in literature to adapt the actual operator tree in a distributed fashion.

We present an optimal distributed algorithm to adapt the placement of a single operator in high communication cost networks, such as a wireless sensor network. Our parameter-free algorithm finds the optimal node to host the operator with minimum communication cost overhead. Three techniques, proposed here, make this feature possible: 1) identifying the special, and most frequent case, where no flooding is needed, otherwise 2) limitation of the neighborhood to be flooded and 3) variable speed flooding and eves-dropping. When no flooding is needed the communication cost overhead for adapting the operator placement is negligible. In addition, our algorithm does not require any extra communication cost while the query is executed. In our experiments we show that for the rest of cases our algorithm saves 30%-85% of the energy compared to previously proposed techniques. To our knowledge this is the first optimal and distributed algorithm to solve the 1-median (Fermat node) problem.

This work aims at minimize the cost of answering snapshot multi-predicate queries in high-communication-cost networks. High-communication-cost (HCC) networks is a family of networks where communicating data is very demanding in resources, for example in wireless sensor networks transmitting data drains the battery life of sensors involved. The important class of multi-predicate queries in horizontally or vertically distributed databases is addressed. We show that minimizing the communication cost for multi-predicate queries is NP-hard and we propose a dynamic programming algorithm to compute the optimal solution for small problem instances. We also propose a low complexity, approximate, heuristic algorithm for solving larger problem instances efficiently and running it on nodes with low computational power (e.g. sensors). Finally, we present a variant of the Fermat point problem where distances between points are minimal paths in a weighted graph, and propose a solution. An extensive experimental evaluation compares the proposed algorithms to the best known technique used to evaluate queries in wireless sensor networks and shows improvement of 10% up to 95%. The low complexity heuristic algorithm is also shown to be scalable and robust to different query characteristics and network size.