Paulo Henrique Portela de Carvalho

A flexible framework based on reinforcement learning for adaptive modulation and coding in OFDM wireless systems

This paper presents a machine learning approach for link adaptation in orthogonal frequency-division multiplexing systems through adaptive modulation and coding. Although machine learning techniques have attracted attention for link adaptation, most of the the schemes proposed so far are based on off-line training algorithms, which make them not well suited for real time operation. The proposed solution, based on the reinforcement learning technique, learns the best modulation and coding scheme for a given signal-to-noise ratio by interacting with the radio channel and it does not rely on an off-line training mode. Simulation results show that under specific conditions, the proposed technique can outperform the well-known solution based on look-up tables for adaptive modulation and coding, and it can potentially adapt itself to distinct characteristics of the radio environment.

A Traffic Analysis per Application in real IP/MPLS Service Provider Network

In this work, real traffic of a service provider IP/MPLS network is classified per application using a traffic characterization procedure based on self-similarity. The results of such classification show that the real traffic confirms a high degree of burstness due to high values of the Hurst parameter. These results are used for the proposal of a method of network capacity planning. The major difficulty is the knowledge of the traffic inside the backbone, and this needs to be a self-learning process mainly because every day new service applications are inserting new type if traffic. In this work an ILEC (incumbent local exchange carrier) service provider IP traffic will be classified per application after that applying a self similarity characterization some analysis will be done.

Network Planning Optimization for Multimedia Networks

This work presents a planning methodology for multimedia networks based on a hybrid traffic model and an evolutionary optimization procedure. The methodology intends to optimize the sizing of network elements to comply with two QoS simultaneous network parameters as well as to promote network stability and cost efficiency. The hybrid traffic model deals with the multimedia network traffic as a combination of fBm (fractional Brownian motion) and Markovian process. Also, the model considers the links loads and variance coefficients to produce more accurate Qos metrics values for network delay and packet loss probability. The traffic model characterization accuracy was evaluated with an optimization procedure that sizes network elements to fulfill performance and cost requirements. The simulations results show that the methodology may have a practical use for sizing network elements and promoting stability.

Internet capacity: optimizing autonomous system inbound traffic using specialist knowledge as support for decision-making

Autonomous systems provide network control without external monitoring. This can be a problem when the traffic is asymmetric with a higher portion being inbound, as networks are generally designed for symmetrical loads. This problem occurs frequently in networks within South American countries as the inbound and outbound portions of traffic are unbalanced by a factor greater than 5:1. Furthermore, national content is usually hosted in the USA, exacerbating the problem. In this study, networks in Brazil, Colombia, and Venezuela are analyzed and characterized. Decision-making techniques are proposed as a tentative external control method or, in some cases, as a proactive attempt to interfere in an internal decision. Our new adaptive algorithm outperformed manual human intervention, which has potential utility to assist long-term capacity planning of the interconnection between autonomous systems.

Multi-objective memetic algorithm applied to the automated synthesis of analog circuits

In this paper, a Multi-objective Memetic Algorithm applied to the automated synthesis of analog circuits is proposed. The optimization of circuit topologies and their parameters are simultaneously carried out. A variable-size 2D circuit representation is used. In this approach, the initial solutions are created based on expert knowledge through the use of well-known buildingblocks and rules-based coupling schemes. The proposed genetic operators are specific to 2D encoding and they are capable of fomenting a balance between diversity and convergence. A local search process – the Simulated Annealing method – is applied in order to improve the circuit parameters. The results show that the proposed method generates, with small populations and few generations, small well-structured circuits which accomplish the specifications.

Integrated Doherty RF CMOS Power Amplifier design for average efficiency enhancement

This paper deals with the implementation of a RF Doherty Power Amplifier (DPA) with the objective of improving the average efficiency. This technique is an interesting way to provide efficient PA for high PAPR signal of more recent standards of communications. The Doherty principle is applied to a 2.5GHz fully integrated PA on a CMOS 65nm technology. The DPA exhibits 23.4dBm output power, 15dB of power gain and 24.7% of PAE on a 7 dB power range. The die size is 2.89mm2. To fulfill high data rates, wide-band behavior is a big challenge. Hence the wideband behavior of integrated DPA is also investigated.

OFDM channel prediction using set-membership affine projection algorithm in time-varying wireless channel

In order to take advantage of adaptive transmission techniques or multiple antennas, it is necessary to have actual channel state information (CSI) at transmitter and/or receiver. Due to system constraints and the time varying nature of the mobile radio channel, the CSI might be outdated, leading to performance degradation of the system. Channel prediction can provide up-to-date channel state information and reduce capacity loss. This paper presents a channel predictor based on the set-membership affine projection filtering involving an OFDM system. Realistic physical channel model used for standardization is considered to evaluate the performance of the predictor and compare it to well-known adaptive algorithms NLMS and RLS. Simulation results also show the bit-error rate performance and the robustness of the proposed predictor under the specifications of 3GPP Long Term Evolution (LTE).

Internet Traffic Engineering: Understanding the Specialist Decision-Making Process for Traffic Management in South America

The importance of the traffic control is a huge paradigm on the Internet. The AS (Autonomous Systems) by definition assumes no external control as priority; this could be an enormous problem when the traffic is asymmetric and the inbound is the higher portion. The problem is very often in networks hosted in South America countries; making the unbalancing inbound and outbound in a rate of more than 5:1 in some situations. In this article some Real South America networks will be analyzed, characterized including the specialist (e.g. Network Engineer) decision-making.

IP and 3G Bandwidth Management Strategies Applied to Capacity Planning

This chapter discusses the application of methodologies to plan and design IP Backbones and 3G access networks for todays Internet world. The recent trend of the multi-frequency band operations for mobile communication systems requires increasingly bandwidth capacity in terms of core and access. The network planning task needs mathematical models to forecast network capacity that match the service demands. As the nature of network usage changed, to explain and forecast the network growth, new methods are needed. In this chapter, we will discuss some strategies to optimize the bandwidth management of a real service provider IP/MPLS backbone and later we will propose a method for traffic engineering in a national IP backbone.

A Planning and Adaptive Routing Methodology for Multimedia Networks

This work presents a traffic engineering proposal based on a network planning and adaptive routing strategy for multimedia networks. The proposal uses a hybrid traffic model based on fBm (fractional Brownian motion) and Markovian processes to characterize the multimedia traffic and calculate QoS metrics and effective bandwidth. The routing strategy was evaluated in a scenario with different Poisson and self-similar aggregated traffic series. The results show that the proposed methodoly achieves the established QoS requirements and promotes a better load balancing in the network.