Mariusz Kamola

Reconstruction of a social network graph from incomplete call detail records

Real-life call detail data (CDR) are used to build a graph of a social network of telecommunication operator customers. Affiliation network is used in graph construction since CDR data are partially kept anonymous. A number of the resulting network properties are examined to prove the correctness of the graph construction algorithm. Cliques in the network and network dynamics are analyzed; suggestions are given about possible utilization of the obtained information in the operation of a telecommunication operator.

Hybrid Approach to Design Optimisation: Preserve Accuracy, Reduce Dimensionality

Hybrid Approach to Design Optimisation: Preserve Accuracy, Reduce Dimensionality The paper proposes a design procedure for the creation of a robust and effective hybrid algorithm, tailored to and capable of carrying out a given design optimisation task. In the course of algorithm creation, a small set of simple optimisation methods is chosen, out of which those performing best will constitute the hybrid algorithm. The simplicity of the method allows implementing ad-hoc modifications if unexpected adverse features of the optimisation problem are found. It is postulated to model a system that is smaller but conceptually equivalent, whose model is much simpler than the original one and can be used freely during algorithm construction. Successful operation of the proposed approach is presented in two case studies (power plant set-point optimisation and waveguide bend shape optimisation). The proposed methodology is intended to be used by those not having much knowledge of the system or modelling technology, but having the basic practice in optimisation. It is designed as a compromise between brute force optimisation and design optimisation preceded by a refined study of the underlying problem. Special attention is paid to cases where simulation failures (regardless of their nature) form big obstacles in the course of the optimisation process.

Network-wide power management in computer networks

An important part of the modern computer networks design is to develop novel technologies, architectures and control mechanisms for network devices enabling power saving by adapting network capacities to current traffic loads and user demands. We describe centralize and hierarchical control frameworks for reducing power consumption in backbone computer networks. The implementation of these frameworks provides the local control mechanisms that are implemented in the network devices level and network-wide control strategies implemented in the central control level. In this paper, we focus on network-wide algorithms for calculating the power status of network devices and the energy-aware MPLS routing for recommended network configuration. We enumerate several possible formulations of a network energy saving optimization problem with continuous and discrete variables. We discuss the limitations of these approaches and problems with their application to power control in real networks. We propose the relaxation of the complete binary problem formulation assuming full routing and energy state of all devices calculation, and the algorithm to solve it. Our formulation is based on a heuristic approach that leads to a continuous optimization. The evaluation of the optimization scheme through simulation is presented in the final part of the paper.

NETWORK RESILIENCE ANALYSIS: REVIEW OF CONCEPTS AND A COUNTRY-LEVEL. CASE STUDY

This paper presents the rationale behind performing an analysis of Internet re- silience in the sense of maintaining a connection of autonomous systems in the presence of failures or attacks — on a level of a single country. Next, the graph of a network is constructed that represents interconnections between autono- mous systems. The connectivity of the graph is examined for cases of link or node failure. Resilience metrics are proposed, focusing on a single autonomous system or on overall network reliability. The process of geographic location of networking infrastructure is presented, leading to an analysis of network resi- lience in the case of a joint failure of neighboring autonomous systems.

Software environment for market balancing mechanisms development, and its application to solving more general problems in parallel way

A new software that supports research on algorithms for resource allocation in multi-commodity markets is presented. Thanks to a much more general data model used, and possibility of plugging in external optimization engines, various solvers can be used to extend the functionality of the platform. The software functionality is focused on supporting researchers in the algorithm engineering process, facilitating e.g. analysis and comparison of market strategies. An example application of the software to travelling salesman problem, solved by many agents, is presented.

Internet of Things with Lightweight Identities Implemented Using DNS DANE—Architecture Proposal

Domain Name Service (DNS) and its certification related resource records are appealing alternative to the standard X.509 certification framework, in provision of identities for Internet of Things (IoT) smart devices. We propose to also use DNS to store device owner identification data in device certificates. A working demonstration software has been developed as proof of this concept, which uses an external identity provider run by national authorities. As a result, smart devices are equipped with certificates that safely identify both the device and its owner. Hardware requirements make such a framework applicable to constrained devices. It stimulates mutual trust in machine-to-machine and man-to-machine communication, and creation of a friendlier environment for sale, lease, and data exchange. Further extensions of the proposed architecture are also discussed.

Using rate equation for modeling triad dynamics on Instagram

Triadic analysis is a convenient way to assess structure and stability of a graph. This paper verifies stability of one triad type that is commonly perceived as transient and unstable, on Instagram subgraph, reconstructed by a specific crawling algorithm. Dynamics of that triad transition has been examined, wrt. degrees of that triad nodes and other basic structural properties of the triad neighborhood. Results show that the triad transition can be modeled as rate equation of order 2, and that it is fairly independent of any of the considered factors. A complete model of dynamics of triads can be further used in graph evolution forecasting, including phenomena like opinion spread or group forming, which may, in turn, affect real-life situations.

Analytics of Industrial Operational Data Inspired by Natural Language Processing

Industrial processes provide lots of operational data on different timescales. Those data are well-structured and used now for daily control, longer-term management and forensics. We propose to pre-process that data and treat them the way the natural language processing is done - first, in order to find common ways the process is controlled. Such knowledge can then be used in prediction or early detection of faults, or necessary manufacturing shifts. Gas transmission operational data are considered here as the live example.

Protecting privacy of GPS trails by anonymization of the road graph

The problem of anonymization of car GPS traces is addressed. Unlike the existing approaches, we propose to preserve anonymity of the whole road graph, while performing accurate GPS location projection onto that graph edges. Projected locations are denoted by their distances from graph vertices, which allows to perform many analytical tasks, in a graph abstracted from physical node locations. With adequately selected subset of samples, the released data will still be enough to perform refined analysis, e.g. of the driving style, while preserving the anonymity of the graph. Possible deanonymization attacks and countermeasures are discussed shortly.

Shared memory for Matlab and Octave: Another package for distributed programming

Abstract The need of distributed computation in math environments as Matlab or Octave is nowadays unquestionable. A parallelisation package is presented that gives the user possibility of distributed programming using common memory model and synchronisation mechanisms, also while programming cross-platform Octave and Matlab. Functionality and technical details of the package are described. Also, a benchmarking test is done for comparison of this package and Matlab Distributed Computing Toolbox efficiency. The test problem is Google PageRank algorithm. The results of tests are promising, but the package still needs further maintenance and development.