Catálogo de publicaciones - libros

It is commonly believed that Ant Routing algorithms cannot be applied with the TCP transport layer. We show that, contrary to this belief, TCP in the transport layer still enables the adaptive algorithms to extend the range of load levels under which they can find efficient routing policies.

A mimicry attack is an exploit in which basic behavioral objectives of a minimalist ’core’ attack are used to design multiple attacks achieving the same objective from the same application. Research in mimicry attacks is valuable in determining and eliminating detector weaknesses. In this work, we provide a process for evolving all components of a mimicry attack relative to the Stide (anomaly) detector under a Traceroute exploit. To do so, feedback from the detector is directly incorporated into the fitness function, thus guiding evolution towards potential blind spots in the detector. Results indicate that we are able to evolve mimicry attacks that reduce the detector anomaly rate from ~67% of the original core exploit, to less than 3%, effectively making the attack indistinguishable from normal behaviors.

We demonstrate how a genetic algorithm solves the problem of minimizing the resources used for network coding, subject to a throughput constraint, in a multicast scenario. A genetic algorithm avoids the computational complexity that makes the problem NP-hard and, for our experiments, greatly improves on sub-optimal solutions of established methods. We compare two different genotype encodings, which tradeoff search space size with fitness landscape, as well as the associated genetic operators. Our finding favors a smaller encoding despite its fewer intermediate solutions and demonstrates the impact of the modularity enforced by genetic operators on the performance of the algorithm.

This paper proposes a novel optimization algorithm, Bacterial Foraging Algorithm with Varying Population (BFAVP), to solve Optimal Power Flow (OPF) problems. Most of the conventional Evolutionary Algorithms (EAs) are based on fixed population evaluation, which does not achieve the full potential of effective search. In this paper, a varying population algorithm is developed from the study of bacterial foraging behavior. This algorithm, for the first time, explores the underlying mechanisms of bacterial chemotaxis, quorum sensing and proliferation, etc., which have been successfully merged into the varying-population frame. The BFAVP algorithm has been applied to the OPF problem and it has been evaluated by simulation studies, which were undertaken on an IEEE 30-bus test system, in comparison with a Particle Swarm Optimizer (PSO) [1].

The Steiner Tree Problem (STP) in graphs is a well-known NP-hard problem. It has regained attention due to the introduction of new telecommunication technologies, since it is the mathematical structure behind multi-cast communications. The goal of this paper is to design an ant algorithm (called ANT-STP) for the STP in graphs which is better than TM, which is a greedy constructive method for the STP proposed in [34]. We derive ANT-STP from TM as follows: each ant is a constructive heuristic close to TM, but the population of ants can collaborate by exchanging information by the use of the trail systems. In addition, the decision rule used by each individual ant is different from the decision rule used in TM. We compare TM and ANT-STP on a set of benchmark problems of the OR-Library.

The main goal of this work is to study the design of message authentication protocols by using cellular automata. Specifically, memory cellular automata with linear transition functions are considered. It is shown that the proposed protocol is secure against different cryptanalytic attacks.

This paper proposes a neural network based approach for solving the resource discovery problem in Peer to Peer (P2P) networks and an Adaptive Global Local Memetic Algorithm (AGLMA) for performing the training of the neural network. This training is very challenging due to the large number of weights and noise caused by the dynamic neural network testing. The AGLMA is a memetic algorithm consisting of an evolutionary framework which adaptively employs two local searchers having different exploration logic and pivot rules. Furthermore, the AGLMA makes an adaptive noise compensation by means of explicit averaging on the fitness values and a dynamic population sizing which aims to follow the necessity of the optimization process. The numerical results demonstrate that the proposed computational intelligence approach leads to an efficient resource discovery strategy and that the AGLMA outperforms two classical resource discovery strategies as well as a popular neural network training algorithm.

In this work, the main goal is to develop and evaluate a number of optimization algorithms in the task of improving levels in TCP/IP based networks, by configuring the routing weights of link-state protocols such as OSPF. Since this is a complex problem, some meta-heuristics from the arena were considered, working over a mathematical model that allows for flexible cost functions, taking into account several measures of the network behavior such as network congestion and end-to-end delays. A number of experiments were performed, resorting to a large set of network topologies, where , and some common heuristic methods including were compared. make the most promising alternative leading to solutions with an effective network performance even under unfavorable scenarios.

Wireless Sensor Networks (WSNs) are becoming an active area of research. They consist of small nodes with limited sensing, computation and wireless communication capabilities. The success of WSNs in real world applications is primarily dependent on a key requirement: ability to provide a communication infra-structure for dissemination of sensed data to a sink node in an energy efficient manner. Therefore, in this paper we propose a bee-inspired power aware routing protocol, , that utilizes a simple bee agent model and requires little processing and network resources. The results of our extensive experiments demonstrate that delivers better performance in a dynamic WSNs scenario as compared to a WSN optimized version of Adhoc On-demand Distance Vector () protocol while its computational and bandwidth requirements are significantly smaller.

Radio Network Design (RND) is a Telecommunications problem that tries to cover a certain geographical area by using the smallest number of radio antennas, and looking for the biggest cover rate. Therefore, it is an important problem, for example, in mobile/cellular technology. RND can be solved by bio-inspired algorithms, among other options, because it is an optimization problem. In this work we use the PBIL (Population-Based Incremental Learning) algorithm, that has been little studied in this field but we have obtained very good results with it. PBIL is based on genetic algorithms and competitive learning (typical in neural networks), being a new population evolution model based on probabilistic models. Due to the high number of configuration parameters of the PBIL, and because we want to test the RND problem with numerous variants, we have used grid computing with BOINC (Berkeley Open Infrastructure for Network Computing). In this way, we have been able to execute thousands of experiments in only several days using around 100 computers at the same time. In this paper we present the most interesting results from our work.