IoTSS: Internet of Things for Smart Systems

Introduction

The Chair of Connected Mobility (I11) recently received an Internet of Things (IoT) Technology Research Award from Google and is looking for research assistants to carry on our proposed project.

The general focus of our Internet of Things for Smart Systems (IoTSS) is the user context, in which way end users react and interact with the presence of an ecosystem consisting of smart devices dedicated to augment their reality and enrich their experience. Therefore, we create a platform composed of different devices able to tackle different research questions spanning across multiple fields:

Indoor navigation optimization
Creation of mobile clouds on resource-constrained devices
Feasibility of virtualization and its advantages on mobile devices
Augmented reality supported by the physical web and IoT devices
Proximity modeling based on multimodal data
Secure and private data handling regarding cloud upload, etc.

We have been granted access to the following hardware and software technologies:

Besides the Google equipment, our setting additionally includes other third-party components called gateways, hosting a plethora of different sensors and able to actively change the surrounding environment (actuators). Thus, end-users will be able to receive context information from the central cloud based on their position as well as in a decentralized manner directly from the gateways. In the aforementioned scenario all components actively participate in providing data. Moreover, the end-users will publish information about themselves to the cloud through their smartphones. This a crucial point especially for privacy related matters.

Research Project Goals

These are the steps and outcomes of our research project:

Proximity Modeling (What’s around us): First, improve indoor navigation like the measurement of received signal strength indication (RSSI) or positioning systems using textual signs with the support of Google beacon platform and the Nearby Message API. Second, enrich the proximity information with further modalities, such as sound and light information for a more reliable detection whether users are still in proximity or not. We want to create a map with users and their surrounding context and infer proximity information by using different metrics like Pearson correlation coefficient.
Privacy: Due to the massive data collection about users by providing useful mobile services, malicious entities can infer user’s whereabouts, interests and social relationships. Thus, we implement several existing privacy approaches like position dummies, anonymity, spatial obfuscation and encryption and evaluate these techniques with a generally applicable framework. For a better privacy, we plan to combine cryptography and obfuscation methods and consider advanced attackers using different types of attacks, such as personal context linking attack and map matching.

Open Positions

Four research assistants initially starting from beginning of June, 2016. The project is meant to last at least 6 months with an expected workload of 8 hours per week. Each topic includes two research assistant positions, they can share knowledge and experience and help each other.

Topic 1: Data Aggregation/Collection and OS Resources Virtualization

Description

This position is oriented towards understanding how virtualization techniques can be exploited to realize the so called Embedded Cloud. With the support of the hardware and software provided by Google we want to create a sensing environment where data are collected, stored and made available for user to access both directly through a direct connection to the OnHub router or through a mobile connection (by connecting to a front-end, like a web service). Moreover it will be necessary to implement a glue layer, a middleware, devoted to the system orchestration and fundamental to collect valuable metrics and statistics.

Required Skills

Back-end development (for data aggregation and system orchestration) - JAVA/SCALA/OCaml/C++
Web and mobile (Android) development tools
Databases (SQL and/or NoSQL)
OS Virtualization Techniques (Docker, Unikernel)
Good knowledge of Linux OS

How to Apply

Interested bachelor or master students send their documents to vittorio.cozzolino (AT) in.tum.de, including

Short letter of motivation
- Why do you want the assistant job
- How the topic fits into your gathered knowledge and experience
Current CV
Bachelor / Master Transcript of records

Topic 2: Feature Modeling for Proximity Detection and Private Communication

Description

First, we want to model the environment by using only one modality, the distance between different people by using the Eddystone Beacons. Second, after creating this solid basis, we enrich the information with further data, like sound, temperature and light sensor data by using Raspberry Pi to improve the accuracy of the localization and create a multimodal content of our environment. An interesting question; which modalities are most appropriate for the detection of nearby people. For instance, similar background noise. To achieve the result, we are using feature modeling in combination with different machine learning techniques in terms of proximity.

Required Skills

Programming of mobile apps Native application for Android, Hybrid app (e.g. PhoneGap) or Web application (platform independent), if all necessary sensor data are available
Privacy-preserving communication protocols for the synchronization mechanism between client and server (cloud)
Experience in feature selection and modeling as part of machine learning
- Different modalities, require different features
- For example, many systems use MFCC features for sound data or accelerometer data as information for activity recognition.

How to Apply

Interested bachelor or master students send their documents to haus (AT) in.tum.de, including

Short letter of motivation
- Why do you want the assistant job
- How the topic fits into your gathered knowledge and experience
Current CV
Bachelor / Master Transcript of records

Contacts

Vittorio Cozzolino: vittorio.cozzolino (AT) in.tum.de

Michael Haus: haus (AT) in.tum.de