Mobile Systems Interest Group Seminars

Past Seminars (2007):

11 December 2007 2pm 6.12
Speaker: Liam McNamara

Title: Content Sharing based on Colocation Prediction in Urban Transport

Abstract:
People living in urban areas spend considerable amounts of time on public transport, for example, commuting to/from work. During these periods, opportunities for inter-personal networking present themselves, as many members of the public now carry electronic devices equipped with Bluetooth technology. Using these devices, individuals can share content (e.g., news and video clips) with fellow travelers that happen to be on the same train or bus. Transferring media content takes time; in order to maximise the chances of successful downloads, users should identify, among colocated people, those who will travel with them for long-enough periods. We propose a user-centric prediction scheme that collects historical colocation information to determine the best data sources. The scheme works on the assumption that people have a high degree of regularity in their movements. We demonstrate the validity of this assumption on a real, large dataset, that consists of traces of people moving in the London Underground. We demonstrate experimentally that, in this scenario, our prediction scheme achieves very good performance, comparing against random choice.

27 November 2007 2pm 6.12
Speaker: Neal Lathia

Title: Trust-Based Distributed Collaborative Filtering

Abstract:
Collaborative filtering is the dominant technique supporting recommender systems. It offers a community-based approach to solving the information overload problem that each individual would otherwise face, by filtering the content for each individual in a unique and personalised manner. One day information filtering will be happen on mobile devices, but before we reach this point, we need to consider the distributed scenario. How can each individual find, cooperate with, and evaluate the surrounding users based on the information they provide? A move away from central servers also calls for a move away from the traditional similarity-based algorithms; we need to incorporate the ideas of trust and reliability into these systems as well. This seminar will recount the story of my PhD to date, and be a chance to discuss the directions that I am planning on taking it from here.

20 November 2007 2pm 6.12
Speaker: Vladimir Dyo

Title: A Node Discovery Service for Partially Mobile Sensor Networks [*]

Abstract:
Wireless Sensor Networks (WSNs) are challenging types of networks where resources can be scarce. In particular, battery is often a very limited resource and the radio interface is the culprit for most of the energy consumption. This makes any discovery of other sensors a difficult task, less cumbersome if sensors are fixed but crucial if (some) sensors start being mobile (such as in wildlife monitoring projects with tagged animals). In this paper we propose a middleware offering node discovery for partially mobile wireless sensor networks, where fixed nodes (sinks), deployed in the environment to monitor the movement of entities, detect those patterns with low power consumption. The approach is based on various machine learning techniques which allows for learning and adapting the wake up strategy of the sinks dynamically. We also report on the evaluation of the approach through simulation and use of real movement traces.

[*] A Node Discovery Service for Partially Mobile Sensor Network. MIDSENS 2007.

13 November 2007 2pm 6.12
Speaker: Anders Lindgren

Title: Seal-2-Seal: A Delay-Tolerant Protocol for Contact Logging in Wildlife Monitoring Sensor Networks

Abstract:
Sensor networks are now enabling the monitoring of various environmental phenomena with more accuracy than the previous labour intensive and less technological solutions. This paper is concerned with the application of opportunistic networking techniques to wildlife monitoring, where the sensors are attached to animals moving in their habitat. We present Seal-2-Seal, a novel protocol for logging of node (i.e., animal) contacts in mobile networks and for dissemination of that information to sinks for further analysis. The protocol utilises an efficient data summary mechanism to reduce the amount of information that needs to be transmitted, thus reducing energy consumption. To evaluate the performance of the protocol, we implemented it for the Contiki operating system on sensor devices and ran simulations based on real-life mobility traces using the Cooja emulator.

16 October 2007 2pm 6.12
Speaker: Neal Lathia

Title: Private Distributed Collaborative Filtering [*]

Abstract:
Collaborative filtering has become an established method to measure users' similarity and to make predictions about their interests. However, prediction accuracy comes at the cost of user's privacy: in order to derive accurate similarity measures, users are required to share their rating history with each other. In this work we propose a new measure of similarity, which achieves comparable prediction accuracy to the Pearson correlation coefficient, and that can successfully be estimated without breaking users' privacy. This novel method works by estimating the number of concordant, discordant and tied pairs of ratings between two users with respect to a shared random set of ratings. In doing so, neither the items rated nor the ratings themselves are disclosed, thus achieving strictly-private collaborative filtering. The technique has been evaluated using the recently released Netflix prize dataset.

[*] Private Distributed Collaborative Filtering. RecSys 2007.

9 October 2007 2pm 6.12
Speaker: Matteo Dell'Amico

Title: SOFIA: Social Filtering for Niche Markets

Abstract:
Digital content production and distribution has radically changed our business models. An unprecedented volume of supply is now on offer, whetted by the demand of millions of users from all over the world. Since users cannot be expected to browse through millions of different items to find what they might like, filtering has become a popular technique to connect supply and demand: *trusted* users are first identified, and their opinions are then used to create recommendations. Users' trustworthiness has so far been measured according to one of the following two criteria: *taste similarity* (i.e., "I trust those who agree with me"), or *social ties* (i.e., "I trust my friends, and the people that my friends trust"). The former criteria aims at identifying competent users, but is subject to abuse by malicious behaviours. The latter aims at detecting *well-intentioned* users, but fails to capture the natural subjectivity of tastes. We argue that, in order to be trusted, users must be *both* well-intentioned and competent. Based on this observation, we propose a novel approach that we call *social filtering*. We describe SOFIA, an algorithm realising this approach, and validate its performance on two real large-scale datasets. We demonstrate that the recommendations produced by SOFIA are both accurate and attack resilient.

2 October 2007 3pm 6.12
Speaker: Bence Pasztor

Title: Opportunistic Mobile Sensor Data Collection with SCAR [*]

Abstract:
Sensors are now embedded in all sorts of devices (such as phones and PDAs) and attached to many moving things such as robots, vehicles and animals. The collection of data from these mobile sensors presents challenges related to the variability of the topology of the sensor network and the need to limit communication (for energy or bandwidth saving). Fortunately, the data collected is often delay tolerant and its delivery to the sinks is, in most cases, not time critical. We have devised SCAR, a context aware opportunistic routing protocol (based on CAR) which allows efficient routing of sensor data to sinks, through selection of best paths by prediction over movement patterns and current battery level of nodes. In this talk I will present the main idea of SCAR and its implementation in the Contiki Operating System. I will also present a short evaluation of the protocol using some initial results from Contiki' s network simulator, COOJA, with mobility traces provided by the ZebraNet Project. We compared the performance (in terms of overhead and delivery ratio) with respect to random choice based dissemination.

[*] Opportunistic Mobile Sensor Data Collection with SCAR. MASS 2007.

11 September 2007 3pm 6.12
Speaker: Daniele Quercia

Title: How mobile devices may select digital content

Abstract:
Whenever mobile users share digital content (eg, photos, videos), one thing that their devices should be able to do is to select content. This talk will look at two problems when selecting content: - The first problem is how to select content that is of high-quality. Consider two devices: A and B. To decide whether to accept content from B, A should somewhat set its trust for B (the extent to which it can rely on B for high-quality content). According to our recent paper[*], A may do so by "learning over graphs". This talk will present that paper. - The second problem is how to select content that is relevant. By considering that mobile users tag content, this talk will touch on the work we are currently doing on how to find the best matching content for a user query.

[*] Lightweight Distributed Trust Propagation. ICDM 2007.

22 June 2007 3pm 6.12
Speaker: Mirco Musolesi

Title: A Connectivity Trace Generator for Testing Opportunistic Mobile Systems

Abstract:
In this talk we discuss the design Connectivity Trace Generator (CTG), a tool for the automatic generation of connectivity traces, which takes as input real mobility traces and is able to output a set of traces with similar connectivity properties, which can be used to test opportunistic mobile systems. This allows developers to investigate the impact of the variation of connectivity patterns, number of hosts, and other parameters on the protocol or application under investigation. We use a real case study (the Dartmouth campus connectivity traces) to show how CTG allows protocol developers to play with some connectivity and density parameters so to best conduct performance testing of different aspects of protocols and applications.

15 June 2007 3pm 6.12 - Reading Group
Organizer: Anders Lindgren

Title: Analysis and Implications of Student Contact Patterns Derived from Campus Schedules

Reading Group:
The paper to be discussed is "Analysis and Implications of Student Contact Patterns Derived from Campus Schedules", Vikram Srinivasan, Mehul Motani, and Wei Tsang Ooi, in Proceedings of MobiCom 2006. The paper can be found at http://www.comp.nus.edu.sg/~ooiwt/papers/contact-mobicom06-final.pdf

8 June 2007 3pm 6.12
Speaker: Vladimir Dyo

Title: Data Collection in Mobile Wireless Sensor Networks

Abstract:
Research in wireless static networks has become mature and there have been a number of successful deployments. However, a new class of applications is emerging where mobility is a major requirement. These applications include wildlife tracking, healthcare and mobile robotics. In this talk, we will discuss the challenges related to energy efficiency in low power mobile sensor networks. One of the major challeges is maintaining network connectivity while putting the nodes to sleep for as much as possible. We will present a scheduling approach which uses a combination of asynchronous wake-up scheduling and detection of temporal regularities in the arrival patterns of encounters. We present our preliminary results and discuss the future work.

7 June 2007 3pm 6.12
Speaker: Ilias Leontiadis

Title: GeOpps: Opportunistic Geographical Routing for Vehicular Networks

Abstract:
Vehicular networks can be seen as an example of hybrid delay tolerant network where a mixture of infostations and vehicles can be used to geographically route the informa- tion messages to the right location. In this paper we present a forwarding protocol which exploits both the opportunis- tic nature and the inherent characteristics of the vehicular network in terms of mobility patterns and encounters, and the geographical information present in navigator systems of vehicles. We also report about our evaluation of the pro- tocol over a simulator using realistic vehicular traces and in comparison with other geographical routing protocols.

7 June 2007 3pm 6.12
Speaker: Mirco Musolesi

Title: TACO-DTN: A Time-aware Content-based Dissemination System for Delay-Tolerant Networks

Abstract:
Content dissemination applications are becoming more and more popular on fixed infrastructure: in this paper we in- troduce TACO-DTN, a content dissemination system which, by virtue of being time-aware in terms of subscriptions and events, is appropriate for delay tolerant networks, where a number of nodes act as infostations, enjoying some form of connectivity to the backbone, and other nodes are mo- bile devices, reachable sometimes only through intermittent connectivity of carriers. Examples of applications benefiting from such a system could be travel information dissemina- tion systems in large cities (exploiting infostations at bus stops) or on highways, advertisements dissemination at spe- cific times, and information dissemination to remote villages. The approach is based on a novel concept of temporal utility of subscriptions and events. The temporal utility is used to govern the routing of the events to the right infostation (i.e., the one reached by the interested subscribers at the right time), avoiding unnecessary information transfer on slow links and the buffer management, in case buffer limi- tations are an issue. We give a description of our protocol and discuss its validation through simulation.

7 June 2007 3pm 6.12
Speaker: Ilias Leontiadis

Title: Opportunistic Spatio-Temporal Dissemination System for Vehicular Networks

Abstract:
Opportunistic dissemination protocols have potentially ap- plications in the domain of vehicular networking, ranging from advertising to emergency/traffic/parking information spreading: one of the characteristics of vehicular networks is that they are often partitioned due to lack of continuity in connectivity among cars or limited coverage of infostations in remote areas. Most available opportunistic, or delay toler- ant, networking protocols, however, fail to take into account the peculiarities of vehicular networks. This paper introduces a novel opportunistic event dissem- ination protocol for vehicular networks. The protocol takes into account the characteristics of these networks in order to dispatch the publications to the subscribers. Furthermore, it uses opportunistic cache and replay mechanisms to deliver the notifications to new subscribers in the area throughout the publication interval. We evaluate our approach through simulation using realistic vehicular traces. We compare our algorithm with a standard epidemic protocol, which offer the best alternative in terms of message delivery, by measuring overhead and delivery over a number of scenarios.

1 June 2007 3pm 6.12
Speaker: Bence Pasztor

Title: Opportunistic Mobile Sensor Data Collection with SCAR

Abstract:
Sensors are now embedded in all sorts of devices (such as phones and PDAs) and attached to many moving things such as robots, vehicles and animals. The collection of data from these mobile sensors presents challenges related to the variability of the topology of the sensor network and the need to limit communication (for energy or bandwidth saving). Fortunately, the data collected is often delay tolerant and its delivery to the sinks is, in most cases, not time critical. We have devised SCAR, a context aware opportunistic routing protocol (based on CAR) which allows efficient routing of sensor data to sinks, through selection of best paths by prediction over movement patterns and current battery level of nodes. In this talk I will present the main idea of SCAR and its implementation in the Contiki Operating System. I will also present a short evaluation of the protocol using some initial results from Contiki’s network simulator, COOJA, with mobility traces provided by the ZebraNet Project. We compared the performance (in terms of overhead and delivery ratio) with respect to random choice based dissemination.

18 May 2007 3pm 6.12
Speaker: Paolo Costa

Title: TeenyLIME: a Tuple Space Middleware for Wireless Sensor Networks

Abstract:
Wireless sensor networks (WSNs) are nowadays evolving from sense-only platforms to complex, sense-and-react systems, where actuators are physically interspersed with the sensors that trigger them. To implement the required application processing, localized interactions must be maximized to improve resource utilization and reduce latency w.r.t. centralized solutions. In this context, application development becomes more complex: the control logic must be embedded inside the network and requires appropriate coordination of the nodes' activities. Moreover, interactions tend to become stateful, and therefore reliable communication acquires an even greater importance. In this talk we present TeenyLIME, a WSN middleware designed to address the aforementioned challenges. TeenyLIME provides application programmers with the high-level abstraction of a tuple space, shared among neighboring WSN nodes. These and other WSN-specific constructs simplify the development of a wide range of applications, including sense-and-react ones. TeenyLIME yields simpler, cleaner, and more reusable implementations, at the cost of only a very limited decrease in performance. We support these claims through a source-level, quantitative comparison between implementations based on TeenyLIME and on alternative approaches, and by analyzing measures of processing overhead and power consumption obtained through cycle-accurate emulation.

11 May 2007 3pm 1.02
Speaker: Vassilis Kostakos

Title: Experiences with Urban Deployment of Bluetooth

Abstract:
As part of the Cityware project, we have spent the last 18 months exploring the use of wireless technologies in urban space. In this talk I will present some of our experiences with Bluetooth technology. Specifically, I will describe some software we have developed that is optimised for capturing a high flow of moving Bluetooth devices in a city context. I will also present some of our findings from deploying Bluetooth "honeypots", which will accept any file sent to them. Finally, I will talk about our ongoing work in studying the diffusion patterns of Bluetooth viruses, the flu, and the Ebola virus in a city.

4 May 2007 3pm 1.02
Speakers: Jon Crowcroft and Ben Hui

Title: Forwarding in small world DTNs in ever decreasing circles

Abstract:
In this talk we seek to improve understanding of the structure of human mobility, and to use this in the design of forwarding algorithms for Delay Tolerant Networks for the dissemination of data amongst mobile users. Cooperation binds but also divides human society into communities. Members of the same community interact with each other preferentially. There is structure in human society. Within society and its communities, individuals have varying popularity. Some people are more popular and interact with more people than others; we may call them hubs. Popularity ranking is one facet of the population. In many physical networks, some nodes are more highly connected to each other than to the rest of the network. The set of such nodes are usually called clusters, communities, cohesive groups or modules. There is structure to social networking. Different metrics can be used such as information flow, Freeman betweenness, closeness and inference power, but for all of them, each node in the network can be assigned a global centrality value. What can be inferred about individual popularity, and the structure of human society from measurements within a network? How can the local and global characteristics of the network be used practically for information dissemination? We present and evaluate a sequence of designs for forwarding algorithms for Pocket Switched Networks, culminating in Bubble, which exploit increasing levels of information about mobility and interaction.

27 April 2007 3pm 6.12
Speaker: Richard Gold

Title: The RUNES Middleware for Networked Embedded Systems

Abstract:
This talk is about the RUNES (Reconfigurable Ubiquitous Networked Embedded Systems) project has a vision to enable the creation of large-scale, widely distributed, heterogeneous networked embedded systems that interoperate and adapt to their environments. The inherent complexity of such systems must be simplified for programmers if the full potential for networked embedded systems is to be realised. The widespread use of network embedded systems requires a standardised architecture which allows self-organisation to suit a changeable environment. RUNES aims to provide an adaptive middleware platform, a common language that will simplify the application creation process. This will allow for a dramatic cut in the cost of new application development and a much faster time to market, transforming applications which are already technically possible into forms that are easy and straightforward for designers to use. I will present the middleware, its various incarnations and some performance measurements.

20 Apr 2007 3pm 6.12
Speaker: Lu Yan

Title: On building reliable and trustworthy mobile systems

Abstract:
Mobile systems touch on a broad array of disciplines, and significant R&D work has been undertaken over recent years on specific applications, mobility models, network connectivity, context-awareness, etc. Nonetheless, all of these smart objects and their applications are to be implemented into our everyday environments. Those systems should exhibit different aspects of reliability and trustworthiness. It is our hypothesis that those aspects are vital, in both technical and business perspectives, to the promise of the ubiquitous future. In this slot I present recent work on building reliable and trustworthy mobile systems, with experiences gained and lessons learned.

9 March 2007 3pm 6.12
Speaker: Vera Cady

Title: Monitoring Elite Athletes – the SESAME Project

Abstract:
During this talk, I will introduce SESAME consortium and present you with our current progress; specifically I will be talking about sensor technologies as applied to human locomotion. The SESAME consortium is a multidisciplinary group formed to investigate the use of wireless sensor-based systems with offline and real-time processing and feedback in enhancing the performance of elite athletes and young athletes who have been identified as having world class potential. The overall goals of the project lie in enhancing performance, improving coach education, and advancing sports science using a range of both hardware and software technologies to achieve this. In so doing, we are planning to build on the extensive experience that exists both within and outside the consortium in the application of sensor systems to human and animal monitoring, and we will seek to advance that knowledge both in terms of outcomes that are specific to sports and in terms of computer science fundamentals.

2 March 2007 3pm 6.12
Speaker: Alexandre Guitton

Title: Work in progress on unidirectional networks

Abstract:
During this talk, I will present research in progress on networks that are not strongly connected. Indeed, we believe that a significant amount of problems in sensor network deployments are caused by connectivity issues. We address the problem of monitoring wide areas, such as forests. In areas of this size, a dense sensor deployment is not feasible due to prohibitive costs. We assume a sparse deployment where sensors have directed antenna attached, pointing towards the next sensor on the path to the gateway. The novelty of this scenario is two-fold: (i) the network is not strongly connected (it is a directed tree) and (ii) the losses on each link are high (e.g., 50% of the packets are lost). The first problem we try to tackle is the reliability of the transmissions. Is it possible to ensure a reliable transmission when there is no way to receive feedback? Is it scalable? The second problem we consider is the energy-efficiency of the transmission. Is it possible to compress the data at the sender, in such a way that the receiver can uncompress most of the partial data received. Our goal is not to design a good compression algorithm, but a way to make the existing algorithms robust to high losses.

23 February 2007 3pm 6.12
Speaker: Elisa Rondini

Title: Distributed Computation in Wireless Ad Hoc Grid Formations with Bandwidth Control

Abstract:
When an emergency occurs within an indoor environment, fast environmental data gathering is essential in order to provide a situation commander with fine-grained information on which she can base appropriate decisions about how best to respond. Sensor networks have the potential to aid in the provision of such information, but suffer from a number of problems. During an incident, many sensors will record essentially the same event from different perspectives whilst, at the same time, each having the capability to generate large amounts of data at a granularity that is not necessarily operationally useful. Given the limitations of wireless networking, the consequence of a number of sensors attempting to transmit such data simultaneously could be congestive collapse at precisely the point at which optimum operational efficiency is most essential; thus, local data reduction is vital. However, sensor nodes are typically specified as a compromise between sensing performance, computational performance, power efficiency, and ergonomic acceptability for wearables. As a result, they may have insufficient computational power on their own to perform the requisite calculations needed to reduce network load. Consequently, in this paper, we explore the convergence of the field of ad hoc sensor networking with that of computational grids, and present initial work on a novel paradigm that takes into account both the computational capabilities of nodes and the local network conditions when distributing tasks. In achieving this, we have selected, adapted and implemented two load sharing algorithms on a real sensor testbed running emulated jobs, and demonstrate that a significant performance improvement, from the point of view of average latency, can be obtained for both the chosen algorithms by taking bandwidth into account.

16 February 2007 3pm 6.12
Speaker:Ettore Ferranti

Title: Brick&Mortar: an on-line multi-agent exploration algorithm

Abstract:
When an emergency occurs within a building, it is critical to explore the area as fast as possible in order to find victims and identify hazards. We propose Brick&Mortar, an algorithm for the autonomous exploration of unknown terrains by a team of mobile nodes, referred to as agents. Because of the unreliability and short range of wireless communications in an indoor environment we suggest that agents communicate indirectly with each other by tagging the environment. Agents have no prior knowledge of the terrain map, but are able to coordinate in order to explore a variety of terrains with different topological features. In our experimental evaluation, we show that Brick&Mortar significantly outperforms the competing algorithms, namely Ants and Multiple Depth First Search, in terms of exploration time. The observed performance benefits suggest that our algorithm is suitable for safety-critical applications that require rapid area coverage for real-time event detection and response.

We will discuss about the paper: Improved Access Point Selection
Anthony J. Nicholson et al. from Mobisys 2006

Abstract:
This paper presents Virgil, an automatic access point discovery and selection system. Unlike existing systems that select access points based entirely on received signal strength, Virgil scans for all available APs at a location, quickly associates to each, and runs a battery of tests to estimate the quality of each APs connection to the Internet. Virgil also probes for blocked or redirected ports, to guide AP selection in favor of preserving application services that are currently in use. Results of our evaluation across five neighborhoods in three cities show Virgil finds a usable connection from 22% to 100% more often than selecting based on signal strength alone. By caching AP test results, Virgil both improves performance and success rate. Our overhead is acceptable and is shown to be faster than manually selecting an AP with Windows XP.

19 January 2007 3pm 6.12
Speaker: Ilias Leontiadis

Title: Opportunistic Dissemination for Vehicular Networking

Abstract:
Opportunistic dissemination protocols could have a potential number of applications in vehicular networking, ranging from advertising to emergency/traffic/parking information spreading: one of the characteristics of vehicular networks is that they are often partitioned due to lack of continuity in connectivity among cars or coverage of infostations in remote areas. However the currently available opportunistic, or delay tolerant, networking protocols, lack some peculiarities which would make them more suitable to vehicular scenarios. For instance, information dissemination applications would benefit from the ability to identify areas in which existing hosts interested to that type of information (subscribers) should be informed, the time interval during which the notification should be valid, and, sometimes, the percentage of subscribers that should receive the notification. We introduce novel opportunistic dissemination protocols, which combine location, time and reliability in its primitives. These protocols take into account the characteristics of vehicular networks in order to dispatch information more efficiently.

12 January 2007, 3pm Room 6.12 MPEB
Reading Group

We will discuss the paper: "CarTel: A Distributed Mobile Sensor Computing System",
Bret Hull, Vladimir Bychkovsky, Yang Zhang, Kevin Chen, Michel Goraczko, Allen K. Miu, Eugene Shih, Hari Balakrishnan, Samuel Madden. In Proceedings of SenSys, Boulder, CO, November 2006.

12 December 2005 12 noon Room 4.02
Reading Group

Title: CarTel: A Distributed Mobile Sensor Computing System

We will discuss the paper "Socially Aware Computation and Communication" (Computer, vol. 38, no. 3, pp. 33-40, March, 2005) by Alex Pentland.
You can download the paper here

This page was last modified 25/06/07