Andreea-Cristina Petre

Energy-Aware Social-Based Routing in Opportunistic Networks

In particular types of Delay-Tolerant Networks (DTN) such as Opportunistic Mobile Networks, node connectivity is transient. For this reason, traditional routing mechanisms are no longer suitable. New approaches use social relations between mobile users as a criterion for the routing process. We argue that in such an approach, nodes with high social popularity may quickly deplete their energy resources - and, therefore, might be unwilling to participate in the routing process. We show that social-based routing algorithms such as BUBBLE Rap are prone to this behavior, and introduce energy awareness as an important criterion in the routing decision. We present experimental results showing that our approach delivers performances similar to BUBBLE Rap, whilst balancing the energy consumption between nodes in the network.

Predicting Encounters in Opportunistic Networks Using Gaussian Process

In particular types of Delay-Tolerant Networks (DTN) such as Opportunistic Mobile Networks, node connectivity is transient, and connections are sparse and small in length. For this reason, traditional routing mechanisms are no longer suitable. Routing algorithms designed for such networks try to maximize the probability of successful message delivery. The most popular approach is to compute the probability of delivering a message using information such as node contacts and location knowledge, thus using past encounters to predict future ones. In this paper we investigate the predictability of human mobility and interactions patterns. We propose the use of supervised learning techniques together with Gaussian process modeling to predict future encounters based on historical patterns of individual nodes. We analyze their accuracy compared to previous prediction techniques, using real-world mobility data traces.

An integrated architecture for future studies in data processing for smart cities

Abstract Data processing for Smart Cities become more challenging, facing with different handling steps: data collection from different heterogeneous sources, processing sometimes in real-time and then delivered to high level services or applications used in Smart Cities. Applications used for intelligent transportation systems, crowd management, water resources management, noise and air pollution management, require different data processing techniques. The main subject of this paper is to propose an architecture for data processing in Smart Cities. The architecture is oriented on the flow of data from the source to the end user. We describe seven steps of data processing: collection of data from heterogeneous sources, data normalization, data brokering, data storage, data analysis, data visualization and decision support systems. We consider two case studies on crowd management in smart cities and on Intelligent Transportation Systems (ITS) and we provide experimental highlights.

Filters for Wi-Fi Generated Crowd Movement Data

Cities represent large groups of people that share a common infrastructure, common social groups and/or common interests. With the development of new technologies current cities aim to become what is known as smart cities, in which all the small details of these large constructs are controlled to better improve the quality of life of its inhabitants. One of the important gears that powers a city is given by traffic, be it vehicular or pedestrian. As such traffic is closely related to all other activities that take place inside of a city. Understanding traffic is still a difficult process as we have to be able to not only measure it in the sense of how many people are using a particular path but also in analyzing where people are going and when, while still maintaining individual privacy. And all this has to be done at a scale that would cover most if not all individuals in a city. With the high increase in smartphones adoption we can reliably assume that a large part of the population in cities are carrying with them, at all times, at least one Wi-Fi enabled device. Because Wi-Fi devices are regularly transmitting signals we can rely on these devices to detect individuals movements unobtrusively without identifying or tracking any particular individual. Special sensors that monitor Wi-Fi frequencies can be placed around a city to gather data that can later be used to identify patterns in the traffic flows. We present a set of filters that can be used to minimize the amount of data needed for processing and without negatively impacting the result or the information that can be extracted from this data. Part of the filters we present can be deployed at the sensor level, making the entire system more scalable, while a different part can be executed before data processing thus enabling real time information extraction and a broader temporal and spatial range for data analysis. Some of these filters are particular to Wi-Fi but some of them can be applied to any detection system.

Enabling Vehicular Data with Distributed Machine Learning

Vehicular Data includes different facts and measurements made over a set of moving vehicles. Most of us use cars or public transportation for our work commute, daily routines and leisure. But, except of our destination, possible time of arrival and what is directly around us, we know very little about the traffic conditions in the city as a whole. Because all roads are connected in a vast network, events in other parts of town can and will directly affect us. The more we know about the traffic inside a city, the better decisions we can make. Vehicular measurements may contain a vast amount of information about the way our cities function. Information that can be used for more than improving our commute, it is indicative of other features of the city like the amount of pollution in different regions. All the information and knowledge we can extract, can be used to directly improve our life. We live in a world where data is constantly generated and we store it and process it at an ever growing rate. Vehicular Data does not stray from this fact and is rapidly growing in size and complexity, with more and more ways to monitoring traffic, either from inside cars or from sensors placed on the road. Smartphones and in-car-computers are now common and they can produce a vast amount of data: it can identify a cars location, destination, current speed and even driving habits. Machine learning is the perfect complement for Big Data, as large data sets can be rendered useless without methods to extract knowledge and information from them. Machine learning, currently a popular research topic, has a large number of algorithms design to achieve this task, of knowledge extraction. Most of these techniques and algorithms can be directly applied to Vehicular Data. In this article we demonstrate how the use of a simple algorithm, k-Nearest Neighbors, can be used to extract valuable information from even a relatively small vehicular data set. Because of the vast size of most of our cities and the number of cars that are on their roads at any time of the day, standard machine learning systems do not manage to process data in a manner that would permit real time use of the extracted information. A solution to this problem is brought by distributed systems and cloud processing. By parallelizing and distributing machine learning algorithms we can use data at its highest potential and with little delay. Here, we show how this can be achieved by distributing the k-Nearest Neighbors machine learning algorithm over MPI. We hope this would motivate the research into other combinations of merging machine learning algorithms with Vehicular Data sets.

Presumably Simple: Monitoring Crowds Using WiFi

Crowd Monitoring is receiving much attention. An increasingly popular technique is to scan for mobile devices, notably smartphones. We take a look at scanning for such devices by recording WiFi packets. Although research on capturing crowd patterns using WiFi detections has been done, there are not many published results when it comes to tracking movements. This is not surprising when realizing that the data provided by WiFi scanners is susceptible to many seemingly erroneous and missed detections, caused by the use of randomized network addresses, overlap between scanners, high variance in WiFi detection ranges, among other sources. In this paper, we investigate various techniques for cleaning up sets of raw detections to sets that can subsequently be used for crowd analytics. To this end, we introduce two different quality metrics to measure the effects of applying the various techniques. We test our approach using a data set collected from 27 WiFi scanners spread across the downtown area of a Dutch city where at that time a 3-day multi-stage festival took place attended by some 130,000 people.

Enabling Mobile Cloud Wide Spread Through an Evolutionary Market-Based Approach

Mobile clouds are an ongoing research topic that has yet to become ubiquitous as the now popular cloud paradigm. This is because of a number of issues with mobile clouds that still need to be addressed such as: incentives, security, privacy, context, data management, usability, and cost benefits. Out of these issues, the most important one that needs to be addressed is the issue of incentives, without which mobile clouds cannot gain enough users for the concept to be useful. Unlike public, company-owned cloud systems, in mobile clouds, the amount of resources or processing power is directly dependent on mobile cloud users that are in the proximity of the individual that requires extra resources. With an increase in the number of mobile cloud users willing to share resources or willing to use the service offered by others, comes an increase in the likeliness that enough mobile-cloud-enabled devices will be available. In this paper, we study incentives for mobile cloud systems and consider as a solution an evolutionary market-based approach to create these incentives. Creating a market for these systems is particularly difficult because of the large number of individuals that need to be involved and their high mobility.

Delay Tolerant Networks for Disaster Scenarios

Disaster and emergency management refers to a range of activities designed to maintain control over crisis situations, providing the rescue and assistance equipment with a framework for helping victims and reducing its impact. The range of activities include prevention, advance warning, early detection, analysis of the problem and assessment of scope, notification of the public and appropriate authorities, mobilization of a response, containment of damage, and relief and medical care for those affected. One of the challenges in emergency scenarios is the fact that communications can be interrupted, cutting the information flow. This lack of communication infrastructure makes an appropriate response to the disaster more challenging, and leads to reduced quality of services experienced by vulnerable civilians. As an example, emergency scenarios with big agglomerations of people or traffic jams following accidents demand a unified communication infrastructure to optimize the response and decision making. This can be overcome using self-configured wireless networks, because they do not require any pre-existing infrastructure to be established, and are easy to deploy and fast to operate. The continuous use of modern smartphones facilitates the accessibility to wireless technologies. However, when incorporating mobile smartphones into disaster assisting networks, the biggest challenge is that such wireless networks need to be specifically designed and used for supporting victims, people and assistance equipment in crisis scenarios. Because of this, in future mobile networks designed for disaster management, there is a need for new architectures and protocols, capable to adapt existing and available wireless technologies for smart data capturing and decision making. This chapter analyses specific challenges and requirements related to supporting communication in such challenged situations. We present an extensive analysis of networking solutions designed to support situations such the ones described.

Chapter 14 - WiFi Tracking of Pedestrian Behavior

Abstract Tracking pedestrian behavior is receiving increasingly more attention. Various techniques have been used so far, yet tracking through WiFi seems to be the most popular one. This popularity comes from the ubiquity of modern smartphones, of which it is known that most have their WiFi enabled all the time. In this chapter we concentrate exclusively on how this WiFi tracking works, and explain its potentials and pitfalls. Special attention is given to the quality of data from WiFi scanning devices, and how this data can, and should be cleaned up before attempts at extracting information from sets of detected devices. As an illustration of the power of WiFi tracking, we also briefly discuss a few recent results from gathering WiFi data from a large event that attracted over 100,00 people spread across three days.

Proximity Graphs for Crowd Movement Sensors

Sensors are now common, they span over different applications, different purposes and some over large geospatial areas. Most data produced by these sensors needs to be linked to the physical location of the sensor itself. By using the location of a sensor we can construct (mathematically) proximity graphs that have the sensors as nodes. These graphs have a wide variety of applications including visualization, packet routing, and spatial data analysis. We consider a sensor network that measures detections of WiFi packets transmitted by devices, such as smartphones. One important feature of sensors is given by the range in which they can gather data. Algorithms that build proximity graphs do not take this radius into account. We present an approach to building proximity graph that takes sensor position and radius as input. Our goal is to construct a graph that contains edges between pairs of sensors that are correlated to crowd movements, reflecting paths that individuals are likely to take. Because we are considering crowd movement, it gives us the unique opportunity to construct graphs that show the connections between sensors using consecutive detections of the same device. We show that our approach is better than ones that are based on the positioning of sensors only.