• Diese Seite als PDF öffnen.

Algebraic Coding for Distributed Storage

Description

Distributed storage systems with a large number of storage commonly rely on MDS codes, such as Reed-Solomon codes, to protect the system from data loss in the event of node failures while keeping the storage overhead low. In the event of such a failure, the reconstruction of failed nodes induces a large amount of network traffic. In recent years several solutions to this problem have been proposed, most notably locally repairable codes and regenerating codes. We investigate the properties of specific subclasses of these codes, as well as the combination of the two properties.
Further, with the increased demand for data privacy, we develop methods for protecting users' data from the eyes of curious servers in distributed storage systems.

Supervisor:

• Diese Seite als PDF öffnen.

CD Entwickliung (Datenarchitektur) und Datenmodellierung

Description

 The Common Data Environment is a central data platform that supports all the real estate management related processes from the phase of the inception of a building all the way through the whole lifecycle of it. It is a common data platform for all the participant applications, resources and actors in the real estate portfolio management for the whole BMW Group. A special focus is on the seamless data integration between the planning and the operation phase of the real estate management.

The CDE comprises of a set of integrated data storage modules that are providing data to the consumer applications in an organized and controlled manner. The structure of the CDE is given by it’s architecture and the Real Estate Data Model, this latter being the backbone of the communication between the different applications in the real estate IT landscape. A high level data model already exists in the Real Estate department, and it is already integrated in several business processes.

The scope of the project is to further develop the existing data model, to a higher level of detail and enrich it with new data objects. In the second part of the project the architecture of the CDE has to be established to support the different integration uses cases between the CDE and the satellite applications that are contributing with data or are consuming data from this central environment.

Supervisor:

Student

• Diese Seite als PDF öffnen.

Regenerating Codes with Locality

Description

Distributed data storage systems need to be able to tolerate the failure of servers without data loss. To maintain this tolerance, newcomers replace failed servers by downloading the required data from the surviving nodes. This repair process can put severe strain on the system and in recent years several coding theoretic measures have been introduced to mitigate this effect. The most promising approaches are codes with locality, where a small number of surviving nodes suffices for repair in most failure events, and regenerating codes, where the amount of data downloaded for repair is minimized.

This thesis will investigate the combination of advanced constructions of codes with locality and regenerating codes, with the goal of designing codes for storage systems that allow for efficient repair with respect to both, the number of servers involved in repair and the the amount of data downloaded.

Supervisor:

Student

• Diese Seite als PDF öffnen.

Lagrange Coded Computation

Description

Topic for "Hauptseminar Digitale Kommunikationssysteme" (MSEI) (https://www.ei.tum.de/index.php?id=5492) NOT for "Seminar on Topics in Communications Engineering" (MSCE).

In distributed computing a user wishes to compute the evaluation of a function in a set of data points by employing multiple worker nodes. Several aspects related to this concept have been studied in recent literature, such as the resilience to stragglers, i.e., worker nodes that do not answer or have a large delay, and the privacy of the data, i.e., the evaluation points. One solution for the function class of polynomials was presented in [1], which can provide both resistance to stragglers and data privacy. The task of the student is to understand the properties and limitations of this framework.

References

• [1] Yu, Qian, et al. "Lagrange coded computing: Optimal design for resiliency, security and privacy." arXiv preprint arXiv:1806.00939 (2018), https://arxiv.org/abs/1806.00939.

Prerequisites

Supervisor:

Student