Content distribution with mobile-to-mobile cooperation is studied. Data is sent to mobile terminals on a long range link then the terminals exchange the content using an appropriate short range wireless technology. Unicasting and multicasting are investigated, both on the long range and short range links. Energy minimization is formulated as an optimization problem for each scenario, and the optimal solutions are determined in closed form. Moreover, the schemes are applied in public safety vehicular networks, where Long Term Evolution (LTE) network is used for the long range link, while IEEE 802.11 p is considered for inter-vehicle collaboration on the short range links. Finally, relay-based multicasting is applied in high speed trains for energy and delay minimization. Results show that cooperative schemes outperform non-cooperative ones and other previous related work in terms of energy and delay savings. Furthermore, practical implementation aspects of the proposed methods are also discussed.

An across-subjects, post-test only design was used in two experiments to assess the impact of interspersing additional math problems (i.e., briefer problems and/or longer problems) among target math problems on students’ persistence when completing computer-delivered math multiplication problems. In Experiment 1, high school students who worked only target problems completed 32% more target problems and worked 22% longer than those who had briefer problems interspersed. Problem completion rates were significantly higher for those who had briefer problems interspersed. These results suggest that altering assignments by interspersing additional, briefer discrete tasks does not always enhance, and in some instances may hinder academic responding. Stimulus preference and within-trial contrast effects provided possible explanations for these results and indicated that interspersing longer problems could, perhaps, cause students to increase persistence. Experiment 2 was designed to replicate Experiment 1 and extend this line of research by investigating the stimulus preference and within-trial contrast hypothesizes. To increase the number of participants and allow for the evaluation of three conditions, college students served as participants for Experiment 2. In Experiment 2, no significant differences among groups (i.e., control group with only target problems, experimental group with brief problems interspersed, and experimental group with long problems interspersed) were found in the amount of time before college students quit working or in their problem completion accuracy levels. Interspersal of the long problems significantly reduced the number of target problems completed. The results failed to support stimulus preference or within-trial contrast theories. Discussion focuses on theoretical and applied implications related to the additive interspersal procedure, the discrete task completion hypothesis, and the delay reduction hypothesis. Applied implications suggest that educators avoid interspersing longer discrete tasks and exercise caution when interspersing brief tasks.

本研究在減低Wimax mesh mode環境中media access階段會造成的delay現象。我們發現當subscriber station (SS)訊務繁忙時，競爭transmission opportunity (TO)以及3-way handshake的過程容易失敗，造成傳輸效率降低，封包的delay也會增加。同時我們也發現封包的延遲還來自minislot的不良配置。因為原始的Wimax mesh mode對於minislot的配置會造成預約過晚的問題，當預約時間被延長，資料就必須無條件的增加等待時間，因此我們將針對這些問題進行改善，降低在訊務繁忙的狀況下，傳輸效率低落的問題。 我們使用兩種方式來改善上述的問題，其一是由Bayer [4]所提出的動態調整holdoff指數(dynamic holdoff exponent)的方式，藉由調整holdoff指數的大小來縮短holdoff時間，以縮短傳輸延遲；另外，我們也提出一個以節點的臨接區域為配置基準(Neighborhood-Based Minislot Allocation, NBMA) 的方法，透過與其相鄰的節點交換訊息，優化minislot的配置。我們用這兩種方式減少IEEE 802.16中媒體存取階段所造成的傳輸延遲。 實驗結果顯示，在網路負載較輕的情況下，NBMA可以有效的將delay降低為原來的七分之一，改善幅度將近85%，抖動率(jitter)的部分亦有20%的改善，並略為提高傳輸效能8%左右；而當網路負載較重時，delay的改善程度仍有40%左右，jitter部分改善了12%，傳輸效能亦有6%的改善。證明了我們的方法確實可行，並且在改善delay方面有顯著的效果。 / IEEE 802.16 mesh network is a new environment of wireless network. It was designed as a self-organized, distributed scheduling, and multi-hop network. However, it is not robust enough to handle a heavy loading environment for lacking of QoS support. Our research is trying to reduce its media access delay, which comes from both TO (transmission opportunity) competition and improper minislot allocation. TO competition will extend the MSH-DSCH (mesh distributed scheduling) interval and slow down the exchanging speed of control message. Improper minislot allocation comes from the distributed scheduling of minislot. When a subscriber station (SS) allocates too much minislot for low-level traffic, it will defer the allocation of other neighbors’ high-level traffic. We use Bayer’s [4] dynamic exponent to reduce holdoff time of SS, and speed up the exchange of control message. On the other hand, we design an “importance factor” (IM-factor) to score the importance of request. Through the exchange of IM-factor, SS and its neighbors will produce a threshold of IM-factor to filter the unimportant requests, and prevent minislot from being assigned too late. In our experiments, the proposed methods can reduce 85% delay and 20% jitter, and increase bandwidth utility by 7%. It shows that our method in reducing transmission delay is pratical and effective.

Nas grandes cidades ao redor do mundo os congestionamentos são problemas bem conhecidos e compartilhados por todos os setores da sociedade. Não é surpreendente, então, que grande parte dos investimentos públicos caminhem na direção de reduzir paulatinamente o esforço rotineiro que a população itinerante faz para chegar ao trabalho ou retornar ao lar, melhorando sua qualidade de vida. É partilhando desse intuito que se encontrou a motivação inicial para o desenvolvimento deste trabalho. Em outro pólo encontra-se a vertente dos procedimentos evolutivos que ao longo da história da vida na Terra foram responsáveis pela emergência espontânea de uma enorme gama de soluções para os mais variados ecossistemas. A tradução disso para ambientes computacionais criou a classe dos algoritmos evolucionários, dentre os quais os Algoritmos Genéticos (AGs), que se destacaram por serem boas heurísticas de busca por conjuntos de parâmetros que resultem em ótimos globais para problemas de engenharia. A aplicação de AGs para otimizações em engenharia de tráfego possui boa base bibliográfica, mas são raras as aplicações reais em meios onde há escassez de dados e de Sistemas Inteligentes de Transporte (do inglês, Intelligent Transportation Systems ITS). Neste trabalho então foi desenvolvido um novo modelo de simulação mesoscópica sobre o qual um AG é aplicado para encontrar planos semafóricos que reduzam atrasos e paradas em sub-redes congestionadas. A ferramenta é simplificada para execução rápida, usando parâmetros normalmente colhidos pela Companhia de Engenharia de Tráfego de São Paulo (CET-SP) em estudos de revisão de temporização semafórica. Ao fim do trabalho, o estudo de caso em uma sub-rede paulistana resultou em reduções da ordem de 30% no nível de atraso e paradas em relação aos valores obtidos com a simulação dos planos anteriores. Vale ressaltar que o espaço de busca foi reduzido ao sub espaço de planos aceitos pela experiência dos especialistas da CET-SP, e mesmo dentro deste escopo, o algoritmo foi bem sucedido ao descartar soluções ruins e fazer emergirem soluções ótimas coerentes. / In large cities throughout the world, high traffic congestion is a well known problem endured by all parts of society. Not surprisingly, a great deal of public investments is made in the direction of reducing the citizens daily efforts for going from home to work and backwards. It is by sharing this intent that the motivation for this thesis was found. Meanwhile, in another section of human knowledge there are concepts revolving the evolutionary procedures that were responsible for the spontaneous emergence of an immeasurable quantity of solutions for adaptation to an almost as greater quantity of ecossystems. When science brought these concepts to computational environments the class of evolutionary algorithms was born, a class embodied by the Genetics Algorithms (in this text referred as AGs, for Algoritmos Genéticos, in portuguese) which are heuristics that stand out as good alternatives for searching parameter settings that result in global optima for engineering problems. There is a good knowledge base regarding the use of AGs for traffic engineering optimizations, but rare are the real implementations in which specialists have to deal with lack of data and the absence of Intelligent Transportation Systems (ITS). Therefore, in this thesis a new mesoscopic simulation model was developed over which an AG is applied for finding signal timing plans that reduce stops and delays in congested subnetworks. The prototyped tool is simplified for quicker execution, using data that is normally collected by the Traffic Engineering Company of São Paulo (CET-SP) in signal timing revision activities. After the development, one of the numerous citys subnetworks was adopted as the case study, for which the prototype found plans that reduced the stops and delays in approximately 30% when compared to the values measured with the simulation of the old plans. It is worth notice that the search space was reduced to the subspace that only contains solutions accepted by the experience of CET-SPs traffic signal specialists, and within this subgroup, the algorithm succeeded in discarding the bad solutions and providing means for the emergence of coherent global optima.

Nas grandes cidades ao redor do mundo os congestionamentos são problemas bem conhecidos e compartilhados por todos os setores da sociedade. Não é surpreendente, então, que grande parte dos investimentos públicos caminhem na direção de reduzir paulatinamente o esforço rotineiro que a população itinerante faz para chegar ao trabalho ou retornar ao lar, melhorando sua qualidade de vida. É partilhando desse intuito que se encontrou a motivação inicial para o desenvolvimento deste trabalho. Em outro pólo encontra-se a vertente dos procedimentos evolutivos que ao longo da história da vida na Terra foram responsáveis pela emergência espontânea de uma enorme gama de soluções para os mais variados ecossistemas. A tradução disso para ambientes computacionais criou a classe dos algoritmos evolucionários, dentre os quais os Algoritmos Genéticos (AGs), que se destacaram por serem boas heurísticas de busca por conjuntos de parâmetros que resultem em ótimos globais para problemas de engenharia. A aplicação de AGs para otimizações em engenharia de tráfego possui boa base bibliográfica, mas são raras as aplicações reais em meios onde há escassez de dados e de Sistemas Inteligentes de Transporte (do inglês, Intelligent Transportation Systems ITS). Neste trabalho então foi desenvolvido um novo modelo de simulação mesoscópica sobre o qual um AG é aplicado para encontrar planos semafóricos que reduzam atrasos e paradas em sub-redes congestionadas. A ferramenta é simplificada para execução rápida, usando parâmetros normalmente colhidos pela Companhia de Engenharia de Tráfego de São Paulo (CET-SP) em estudos de revisão de temporização semafórica. Ao fim do trabalho, o estudo de caso em uma sub-rede paulistana resultou em reduções da ordem de 30% no nível de atraso e paradas em relação aos valores obtidos com a simulação dos planos anteriores. Vale ressaltar que o espaço de busca foi reduzido ao sub espaço de planos aceitos pela experiência dos especialistas da CET-SP, e mesmo dentro deste escopo, o algoritmo foi bem sucedido ao descartar soluções ruins e fazer emergirem soluções ótimas coerentes. / In large cities throughout the world, high traffic congestion is a well known problem endured by all parts of society. Not surprisingly, a great deal of public investments is made in the direction of reducing the citizens daily efforts for going from home to work and backwards. It is by sharing this intent that the motivation for this thesis was found. Meanwhile, in another section of human knowledge there are concepts revolving the evolutionary procedures that were responsible for the spontaneous emergence of an immeasurable quantity of solutions for adaptation to an almost as greater quantity of ecossystems. When science brought these concepts to computational environments the class of evolutionary algorithms was born, a class embodied by the Genetics Algorithms (in this text referred as AGs, for Algoritmos Genéticos, in portuguese) which are heuristics that stand out as good alternatives for searching parameter settings that result in global optima for engineering problems. There is a good knowledge base regarding the use of AGs for traffic engineering optimizations, but rare are the real implementations in which specialists have to deal with lack of data and the absence of Intelligent Transportation Systems (ITS). Therefore, in this thesis a new mesoscopic simulation model was developed over which an AG is applied for finding signal timing plans that reduce stops and delays in congested subnetworks. The prototyped tool is simplified for quicker execution, using data that is normally collected by the Traffic Engineering Company of São Paulo (CET-SP) in signal timing revision activities. After the development, one of the numerous citys subnetworks was adopted as the case study, for which the prototype found plans that reduced the stops and delays in approximately 30% when compared to the values measured with the simulation of the old plans. It is worth notice that the search space was reduced to the subspace that only contains solutions accepted by the experience of CET-SPs traffic signal specialists, and within this subgroup, the algorithm succeeded in discarding the bad solutions and providing means for the emergence of coherent global optima.

Made available in DSpace on 2016-04-29T13:18:02Z (GMT). No. of bitstreams: 1 Paulo A B Panetta.pdf: 2725022 bytes, checksum: 647e2f895cbc2bf81fd00346a8f0e9ab (MD5) Previous issue date: 2007-05-07 / Fifteen rats were randomized to one of the five following groups (with three rats in each group) to examine the effects of different durations of light ( delay signal ) upon the response acquisition and maintenance of lever press with delayed reinforcement, on the absence of shaping. Two phases construct the experiment, with fifteen experimental sessions and two sessions in extinction on each phase. For the rats randomized to Chain II Group, when Phase 1 was initiated, the duration of the light was initially the same of the delay, but was reduced gradually across phase until it was no longer present with the delay period. For rats randomized to Chain III Group, there was no delay signal initially, but its duration was increased across phase until it equalized the duration of the programmed delay (4 s). The duration of the delay signal was always the same of the delay for the rats employed on Chain I Group, while the was no delay signal for rats randomized to Tand Group. With the rats randomized to the Control Group there was no delay on reinforcement. Within subjects, response rates where higher as the duration of the delay signal increased, with the highest acceleration when its duration was minimal. Response rates where higher also to prior rates when the duration of the delay signal was reduced, with the highest acceleration when its duration was minimal. The findings are discussed in terms of the conditioned reinforcement of the delay signal by reduction the delay between its onset and the reinforcer / Quinze ratos foram distribuídos em cinco grupos com o objetivo de avaliar os efeitos de diferentes durações de luz, que eram apresentadas durante o período de atraso, sobre a aquisição e manutenção da resposta de pressão à barra com atraso de reforço, sem modelagem dessa resposta. O estudo foi dividido em duas fases com dezessete sessões cada (quinze sessões experimentais e duas em extinção). Na Fase 1, para os sujeitos do Grupo Encadeado III, a luz não estava presente nas sessões iniciais. Com o decorrer das sessões, a duração da luz foi aumentada gradualmente até que a sua duração cobrisse todo o período de atraso. Para os sujeitos do Grupo Encadeado II, duração da luz equivalia à totalidade do período de atraso nas sessões iniciais, sendo reduzida gradualmente a sua duração ao longo das sessões até não haver luz durante o período. Nas sessões dos sujeitos do Grupo Tandem não havia luz durante o período de atraso, enquanto nas sessões dos sujeitos do Grupo Encadeado I a duração luz sempre cobria a totalidade do período. Já nas sessões dos sujeitos do Grupo Controle não havia atraso de reforço. Decorrida a Fase 1, o período de atraso foi aumentado de 4 s para 8 s e iniciou-se a Fase 2, idêntica à anterior. Foi observada, em ambas as fases, um aumento na taxa de emissões e de reforços produzidos pelos sujeitos do Grupo Encadeado III na medida que a duração da luz era aumentada, com um aumento maior em ambas as fases quando a luz tinha sua duração mínima. Do mesmo modo, foi notado um pequeno aumento na taxa de emissões e de reforços produzidos pelos sujeitos do Grupo Encadeado II a partir da redução na duração da luz, com aumento maior quando a duração da luz era mínima. Os resultados do presente estudo foram discutidos a partir da noção de reforçador condicionado e da hipótese da redução do atraso proposta por Fantino (1969, 1977)