Dagstuhl Seminar 10042 Demetris Zeinalipour University of Cyprus

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 “Continuous All k Nearest Neighbor Queries in Smartphone Networks” Georgios Chatzimilioudis Demetrios Zeinalipour-Yazti Wang-Chien Lee Marios D. Dikaiakos Tuesday, July 24, 2012 13 th IEEE Int. Conference on Mobile Data Management (MDM’ 12), Bangalore, India MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 1

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Smartphones • Smartphone: A powerful sensing device! – – • Processing: 1. 4 GHz quad core (Samsung Exynos) RAM & Flash Storage: 2 GB & 48 GB, respectively Networking: Wi. Fi, 3 G (Mbps) / 4 G (100 Mbps– 1 Gbps) Sensing: Proximity, Ambient Light, Accelerometer, Microphone, Geographic Coordinates based on AGPS (fine), Wi. Fi or Cellular Towers (coarse). In-House Applications! Smart. Trace (ICDE’ 09, MDM’ 09, TKDE’ 12) Smart. P 2 P (MDM’ 11, MDM’ 12) Airplace (Mobi. Sys’ 12, MDM’ 12) MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 2

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Motivation • Crowdsourcing with Smartphones – A smartphone crowd is constantly moving and sensing providing large amounts of opportunistic data that enables new services and applications "Crowdsourcing with Smartphones", Georgios Chatzimiloudis, Andreas Konstantinides, Christos Laoudias, Demetrios Zeinalipour-Yazti, IEEE Internet Computing, Special Issue: Crowdsourcing (Sep/Oct 2012), accepted May 2012. IEEE Press, 2012 MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 3

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Continuous k Nearest Neighbor Queries Find 2 Closest Neighbors for 1 User MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Motivation “Create a Framework for Efficient Proximity Interactions” Screenshots from a prototype system we've implemented for Windows Phone http: //www. zegathem. com/ MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 5

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Continuous All k Nearest Neighbor Queries Find 2 Closest Neighbors for ALL User MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Presentation Outline • Motivation • System Model and Problem Formulation • Proximity Algorithm • Experimental Evaluation • Future Work MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 7

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 System Model • A set of users moving in the plane of a region (u 1, u 2, … un) • Area covered by a set of Network Connectivity Points (NCP) – Each NCP creates the notion of a cell – W. l. o. g. , let the cell be represented by a circular area with an arbitrary radius • A mobile user u is serviced at any given time point by one NCP. • There is some centralized service, denoted as QP (Query Processor), which is aware of the coverage of each NCP. • Each user u reports its positional information to QP regularly Query Processor u 3 QP u 2 u 0 . u 6 u 5 . . C . MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos . . u. u 1 4 . u 7 8

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Problem Definition • Definition of the CAk. NN problem: Given a set U of n users and their location reports ri, t ∈ R at timestep t ∈ T , then the CAk. NN problem is to find in each timestep t ∈ T and for each user ui ∈ U the k objects Usol ⊆ U − ui such that for all other objects uo ∈ U − Usol − ui, dist(uk, ui) ≤ dist(uo, ui) holds Query Processor u 3 . . . u u 2 u 0 u 6 u 5 . . C 1 4 . u 7 MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 9

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Naïve Solution / Related Work Find 2 -NN for u 0 at timestep t. Query Processor u 3. u 2. u 0. u 6. u 5. C For u 5? Look inside your cell! WRONG! . u 4. u 1 (u 1 closer) TOO EXPENSIVE! . u 7 Perform iterative deepening! MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 10

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Related Work • Existing algorithms for Ck. NN (not CAk. NN) Yu et al. (YPK) 11 and Mouratidis et al. (CPM) 12 • Stateless Version: Iteratively expand the search space for each user into neighboring cells to find the k. NN (like previous slide) • Stateful Version: Improve the stateless version by utilizing previous state, under the following ate ri assumptions: op r pp NN t A Ak No r C fo – i) Static Querying User (i. e. , designated for point queries) – ii) Target users move slowly (i. e. , state does not decay) – iii) Few Target users 11. Yu, Pu, Koudas. “Monitoring k-nearest neighbor queries over moving objects, ” ICDE ’ 05 12. Mouratidis, Papadias, Hadjieleftheriou, “Conceptual partitioning: an efficient method for continuous nearest neighbor monitoring, ” SIGMOD ’ 05. MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 11

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Presentation Outline • Motivation • System Model and Problem Formulation • Proximity Algorithm • Experimental Evaluation • Future Work MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 12

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity Overview • The first specialized algorithm for Continuous All k-NN (CAk. NN) queries Important characteristics: • – – – • Stateless (i. e. , optimized for high mobility) Batch processing (i. e. , with search space searching) Parameter-free (i. e. , no tuning parameters) Generic operator for proximity-based queries – See Crowdcast application presented later. MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 13

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity Outline • For every timestep: 1. Initialize a k+-heap for every cell 2. Insert every user’s location report to every k+-heap • Notice that k+-heap is a heap-based structure and most location reports will be dropped as a result of an insert operation 3. For every user scan the k+-heap of his cell to find his k-NN Query Processor u 3 . u 2 u 0 . u 6 u 5 . . C . . u. u 1 4 . u 7 MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 14

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Intuition behind Proximity • • Users in a cell will share the same search space (search space sharing) Compute 1 search space per cell only! Query Processor u 3 . u 2 Cell Search Space (contains the right answers for all users in C) u 6 u 5 . . u 0 C . . . u. u 4 1 . u 7 MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 15

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity k+-heap • The k+-heap structure for a cell k+-heap structure U list All users inside the cell K top-k heap B ordered list K nearest users Beyond k nearest outside the cell outside users (correctness) MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 16

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 k+-heap Construction (for Cell C) k+-heap structure Assume k=2. Arriving Structure Reports U K Structure B u 6 u 4 u 7 u 6 u 7, u 4 u 2 u 6 u 4, u 2 u 7 u 3 u 6 u 3, u 2 u 4, u 7 u 1 u 6 u 2, u 1 u 3, u 4 u 5 u 6 u 2, u 1 u 3, u 4, u 5 u 0 u 6, u 0 u 2, u 1 u 3, u 4, u 5 U list K top-k heap u 3 . u 2 u 0 . u 6 u 5 . C . MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos B ordered list . . u. u 1 4 . u 7 17

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 k+-heap Construction (for Cell C) k+-heap structure Assume k=2. Arriving Structure Reports U K Structure B u 6 u 4 u 7 u 6 u 7, u 4 u 2 u 6 u 4, u 2 u 7 u 3 u 6 u 3, u 2 u 4, u 7 u 1 u 6 u 2, u 1 u 3, u 4 u 5 u 6 u 2, u 1 u 3, u 4, u 5 u 0 u 6, u 0 u 2, u 1 u 3, u 4, u 5 U list K top-k heap u 3 . . u 2 d r+ te e u 0 m a Di u 6 . . C . . x u 5 MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos B ordered list . u k . u 1 4 . u 7 2 nearest users to cell 18

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Presentation Outline • Motivation • System Model and Problem Formulation • Proximity Algorithm • Experimental Evaluation • Future Work MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 19

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Experimental Methodology • Dataset: – Oldenburg Dataset: • • – Manhattan Dataset: • • • Vehicular mobility generator with roadmap of Manhattan as input (3 km x 3 km) 500 users Network connectivity points (NCPs) uniformly in space – • • Spatiotemporal generator with roadmap of Oldenburg as input (25 km x 25 km) 5000 maximum users (Oldenburg population: 160. 000) Communication ranges 1 km, 4 km and 16 km for the Oldenburg dataset and ranges 1 km and 4 km for the Manhattan dataset Query: CAk. NN Comparison: Proximity vs YPK vs CPM (using the optimal parameter value for YPK and CPM) MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 20

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity - Experiments • • Bottleneck for Proximity: build time Bottleneck for the adapted YPK and CPM: search time MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 21

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity - Experiments • • Benefit of Proximity in the search time Proximity is scalable with respect to k – Search space for Proximity is not proportional to k like YPK, CPM MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 22

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity - Experiments • • Proximity: build time drops as k increases! Machine’s memory scheduler makes more efficient use of buffers when the search spaces are larger MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 23

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity - Experiments • • • Proximity scales with the number of users Proximity outperforms YPK, CPM by an order of magnitude Proximity does not converge to YPK, CPM for higher values of Nmax. MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 24

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 The Crowdcast Application Suite Crowdcast Suite … PROXIMITY WIN 7 API Ranked 3 rd in Microsoft’s Imagine. Cup contest local competition (to appear in the next demo session!) MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Crowdcast: Msg. Cast - Location-based Chat Channel (Post / Follow) - Provide Guidelines "Get your location-based questions answered!" MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Crowdcast: Help. Cast "The Ubiquitous Help Platform for Anyone in Need” MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Crowdcast: Eye. Cast - Disaster Recovery Operations - Indoor Video Conferencing Network "Extend your Vision beyond your 2 eyes!" MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Presentation Outline • Motivation • System Model and Problem Formulation • Proximity Algorithm • Experimental Evaluation • Future Work MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 29

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Proximity – Future Work • Privacy extensions (e. g. , spatial cloaking, strong privacy) • Parallelizing server computation • Proximity for cloud server • User-defined k values MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 30

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Future Work: Smart. Lab. cs. ucy. ac. cy Programming cloud for the development of smartphone network applications & protocols as well as experimentation with real smartphone devices. Install APK, Upload File, Reboot, Screenshots, Monkey Runners, etc. … "Demo: A Programming Cloud of Smartphones", A. Konstantinidis, C. Costa, G. Larkou and D. Zeinalipour-Yazti, "Demo at the 10 th International Conference on Mobile Systems Applications and Services" (Mobisys '12), Low Wood Bay Lake District UK, 2012 MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 31

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 Future Work: Smart. Lab MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 32

Dagstuhl Seminar 10042, Demetris Zeinalipour, University of Cyprus, 26/1/2010 “Continuous All k Nearest Neighbor Queries in Smartphone Networks” Thanks! Questions? Georgios Chatzimilioudis Demetrios Zeinalipour-Yazti Wang-Chien Lee Marios D. Dikaiakos Tuesday, July 24, 2012 13 th IEEE Int. Conference on Mobile Data Management (MDM’ 12), Bangalore, India MDM 2012 © Chatzimilioudis, Zeinalipour-Yazti, Lee, Dikaiakos 33