Bruno R. Andriamanalimanana

Domain-size constraint on real-time model abstractions

Network domain is predicated by the visibility of active nodes from a controller or an observer. The events shaping network factors may affect observation considerably. Accordingly, owing to network congestion and sudden change in resource availability, causal events may lose causal polarity and event bundles may appear slack at the observation post. These nodes are then beyond observation and control. Even though they may appear participating like any other regular nodes, their presence may affect real-time model abstraction processes. Highly dense domains may generate model change points at a faster rate than the observer can process affecting the model abstraction process considerably. In this paper, a framework is explored to articulate the manifold event possibilities that constrain the node visibility, and hence, the domain size. A sketchy optimization model is attempted to realize a limitation of the model abstraction process as a function of hop count.

Model abstractions for real-time network environments

The model abstraction problem is explored from a real-time network environment perspective, usually admitting different system models (such as queue-theoretic) at its different equilibrium states. To computationally depict system states at any level of abstraction, it is necessary to identify correct models consistent with observables. However, any such system identification need not be permanent, particularly for a dynamic system. In such situations, as the system appears to migrate from one equilibrium state to another, one should be able to quickly identify an event of context-switching from one model abstraction to another. In this paper we show how, using a variation of traditional CUSUM statistical approaches, one could identify model change events on time.

Semi-synchronocity Enabling Protocol and Pulsed Injection Protocol For A Distributed Ledger System

Distributed ledger technologies have a central problem that involves the latency. When transactions are to be accepted in the ledger, latency is incurred due to transaction processing and verification. For efficient systems, high latency should be avoided for the governance of the ledger. To help reduce latency, we offer a distributed ledger architecture, Tango, that mimics the Iota-tangle design as articulated by Popov [1] in his seminal paper. We introduce a semi-synchronous transaction entry protocol layer to avoid asynchronism in the system since an asynchronous system has a high latency. We further model periodic pulsed injections into the evaluation layer from the entry layer to regulate the performance of the system.

Trust Management in Distributed Systems via SPATE

This paper describes a peer-to-peer distributed system predicated by a trust management entity that provides a functionally stable, scalable, anonymous peer-centric conglomeration we call SPATE. SPATE stands for soft-profiling architecture for trust evaluation in which a peer is evaluated minimally within a flexible range. If the range remains invariant within a time-span, evaluation scores need not be updated. This results in less update traffics. In this scheme, the system attempts to stay robust against malicious peers with pertinent trust-records replicated and distributed among coordinators.

Unitals and ovals of symmetric block designs in LDPC and space-time coding

An approach to the design of LDPC (low density parity check) error-correction and space-time modulation codes involves starting with known mathematical and combinatorial structures, and deriving code properties from structure properties. This paper reports on an investigation of unital and oval configurations within generic symmetric combinatorial designs, not just classical projective planes, as the underlying structure for classes of space-time LDPC outer codes. Of particular interest are the encoding and iterative (sum-product) decoding gains that these codes may provide. Various small-length cases have been numerically implemented in Java and Matlab for a number of channel models.

Irreducible root systems for space-time coding in wireless communication

The theory of groups and group representations have been shown by several researchers to generate promising space-time codes for multiple-antenna links in wireless communication. In this report, potential codes from a root system in a real Euclidean space are examined. In particular, general properties of codes from the infinite irreducible systems Al, Bl, Cl, and Dl are analytically studied, while codes from the exceptional systems G2, F4, E6, E7 and E8 are numerically determined. Some of these exceptional codes meet general coding-theoretic expectations, for example in terms of the number of code words. The effort described in this paper is part of a general attempt to assess the use of modular representations of groups, as opposed to classical, i.e. complex-numbered, representations, for antenna diversity. Java technologies are used for the implementation of our numerical experimentation platform.

Unitary geometry numerical modeling for wireless communication diversity schemes

Antenna diversity promises to provide high data rates with low error probabilities in wireless communication. An important step for realizing these promises centers around the design of efficient space-time codes for multiple-antenna links. The case of mobile communication is particularly challenging. The use of complex representations of finite algebraic groups has recently been shown by some researchers to be a promising avenue for the design of space-time codes. This paper provides some background on the interplay between group representation theory and wireless communication. Then, it examines the potential contributions of finite groups of Lie type and finite unitary geometry to space-time coding. Java technologies are used for the implementation of our numerical experimentation platform.

Problem signatures from enhanced vector autoregressive modeling

The work reported in this paper concerns the enhancement of mutivariate autoregressive (AR) models with geometric shape analysis data and stochastic causal relations. The study aims at producing numerical signatures characterizing operating problems, from multivariate time series of data collected in an application and operating environment domain. Since the information content of an AR model does not appear sufficient to characterize observed vector values fully, both geometric and stochastic modeling techniques are applied to refine causal inferences further. The specific application domain used for this study is real-time network traffic monitoring. However, other domains utilizing vector models might benefit as well. A partial Java implementation is being used for experimentation.