Structuring communication software for quality-of-serviceguarantees

We propose architectural mechanisms for structuring host communication software to provide QoS guarantees. We present and evaluate a QoS sensitive communication subsystem architecture for end hosts that provides real time communication support for generic network hardware. This architecture provides services for managing communication resources for guaranteed QoS (real time) connections, such as admission control, traffic enforcement, buffer management, and CPU and link scheduling. The architecture design is based on three key goals: maintenance of QoS guarantees on a per connection basis, overload protection between established connections, and fairness in delivered performance to best effort traffic. Using this architecture we implement real time channels, a paradigm for real time communication services in packet switched networks. The proposed architecture features a process per channel model that associates a channel handler with each established channel. The model employed for handler execution is one of “cooperative” preemption, where an executing handler yields the CPU to a waiting higher priority handler at well defined preemption points. The architecture provides several configurable policies for protocol processing and overload protection. We present extensions to the admission control procedure for real time channels to account for cooperative preemption and overlap between protocol processing and link transmission at a sending host. We evaluate the implementation to demonstrate the efficacy with which the architecture maintains QoS guarantees on outgoing traffic while adhering to the stated design goals     

A comparative analysis of methods for pruning decision trees

In this paper, we address the problem of retrospectively pruning decision trees induced from data, according to a top-down approach. This problem has received considerable attention in the areas of pattern recognition and machine learning, and many distinct methods have been proposed in literature. We make a comparative study of six well-known pruning methods with the aim of understanding their theoretical foundations, their computational complexity, and the strengths and weaknesses of their formulation. Comments on the characteristics of each method are empirically supported. In particular, a wide experimentation performed on several data sets leads us to opposite conclusions on the predictive accuracy of simplified trees from some drawn in the literature. We attribute this divergence to differences in experimental designs. Finally, we prove and make use of a property of the reduced error pruning method to obtain an objective evaluation of the tendency to overprune/underprune observed in each method     

Mathematical foundations for error estimation in numericalsolutions of integral equations in electromagnetics

The problem of error estimation in the numerical solution of integral equations that arise in electromagnetics is addressed. The direct method (Green's theorem or field approach) and the indirect method (layer ansatz or source approach) lead to well-known integral equations both of the first kind [electric field integral equations (EFIE)] and the second kind [magnetic field integral equations (MFIE)]. These equations are analyzed systematically in terms of the mapping properties of the integral operators. It is shown how the assumption that field quantities have finite energy leads naturally to describing the mapping properties in appropriate Sobolev spaces. These function spaces are demystified through simple examples which also are used to demonstrate the importance of knowing in which space the given data lives and in which space the solution should be sought. It is further shown how the method of moments (or Galerkin method) is formulated in these function spaces and how residual error can be used to estimate actual error in these spaces. The condition number of all of the impedance matrices that result from discretizing the integral equations, including first kind equations, is shown to be bounded when the elements are computed appropriately. Finally, the consequences of carrying out all computations in the space of square integrable functions, a particularly friendly Sobolev space, are explained     

Tensile and flexural creep rupture tests on partially-damaged concrete specimens

Three series of novel tensile and flexural creep tests on partially-damaged concrete specimens were carried out in order to gain some insight into creep crack growth and failure of strain-softening materials. In the tests, each specimen was initially loaded to a given point in the descending branch and thus had a lower load-carrying capacity than that at the peak-point. Then, the specimen was unloaded and reloaded to sustain a load which was from 70% to 95% of its current load-carrying capacity. Experimental creep curves display a three-stage process, consisting of primary, secondary and tertiary stages, with a decreasing, constant and increasing creep rate, respectively. The secondary stage dominates the whole failure lifetime, whereas both the secondary and tertiary stages are important in terms of creep deformation. Failure life-time seems to be more sensitive to the change of load level in flexural tests rather than in tensile tests. The decrease in load-carrying capacity due to damage tends to result in a shorter failure lifetime and a lower critical load level for creep rupture. The descending branch of the static load-deflection or load-CMOD curve may be used as an envelope criterion for creep fracture.     

Anatomical basis of the variable aspects of injuries of the axillary nerve (excluding the terminal branches in the deltoid muscle)

The course of the axillary n. is complex with three points of angulation that may be used to delineate four segments and a fifth segment that corresponds to the intramuscular ending of the nerve in the deltoid m. The purpose of this study was to determine the precise anatomy of the nerve and of its branches, and some morphologic features for each segment. Thirty-two shoulders from embalmed adult cadavers have been studied. The axillary n. was divided in five segments: 1) from its origin to the inferior border of the subscapularis m., 2) from the subscapularis m. to the anterolateral border of the tendon of the long head of the triceps brachii m., 3) from the triceps to the posteromedial part of the surgical neck of the humerus, 4) from the humerus to the entry into the deltoid m., 5) the intramuscular distribution of the nerve in the deltoid m. In each segment from 1 to 4 were noted the origins of the branches to the subscapularis and teres minor mm. and to the scapulohumeral joint, and the origins of the lateral cutaneous branchial n. and of the terminal motor branches to the deltoid m. The length and the diameter of the nerve in the segments and the distance from the segment S1 to the musculotendinous junction of the subscapularis m. were measured. The results showed that the mean diameters were about 4.1 mm in segment 1, 4.1 mm in segment 2 and 3.4 mm in segment 3. The mean distance to the musculotendinous junction was 7.7 mm. Many variations in the levels of origin of the different muscular, articular or cutaneous branches were found without symmetry between the right and left sides. The lateral cutaneous brachial n. was absent in four cases. The results are compared with those in the literature. The division into five segments is proposed to radiologists and surgeons for evaluation or operative procedures on the axillary n., and to provide a hypothesis about the variable aspects of injuries of the nerve.     

Reminiscence: diffraction and the mountains of Alaska

Discusses, from a personal point of view, experiments carried out to study the diffraction and propagation of VHF, HF and MF radio waves in the early-mid 1950s in the mountains of Alaska     

Relationship between the genomic organization and the overlapping embryonic expression patterns of the zebrafish dlx genes

To understand the relationship between the expression and the genomic organization of the zebrafish dlx genes, we have determined the genomic structure of the dlx2 and dlx4 loci. This led to the identification of the zebrafish dlx1 and dlx6 genes, which are closely linked to dlx2 and dlx4, respectively. Therefore, the inverted convergent configuration of Dlx genes is conserved among vertebrates. Analysis of the expression patterns of dlx1 and dlx6 showed striking similarities to those of dlx2 and dlx4, respectively, the genes to which they are linked. Furthermore, the expression patterns of dlx3 and dlx7, which likely constitute a third pair of convergently transcribed genes, are indistinguishable. Thus, the overlapping expression patterns of linked Dlx genes during embryonic development suggest that they share cis-acting sequences that control their spatiotemporal expression. The evolutionary conservation of the genomic organization and combinatorial expression of Dlx genes in distantly related vertebrates suggest tight control mechanisms that are essential for their function during development.     

High affinity presentation of an autoantigenic peptide in type I diabetes by an HLA class II protein encoded in a haplotype protecting from disease

Tissue composition and the distribution of body mass are described for four genera of East African Bovidae (Madoqua, Gazella, Damaliscus, Hippotragus) with supporting data from four others (Neotragus, Oryx, Tragelaphus, Connochaetes). These species are high in muscle mass, an adaptation convergent with other high-speed terrestrial cursors, bounders, and saltators. The segments below the elbow/cubitus and knee/stifle/genu joints in small bovids are both lighter in percent of total body mass (8.6% TBM) and less heavily muscled (10-15% of total limb musculature) than those segments in macaques (13.6% TBM, 20-25% of the limb musculature). Bovid species differ from one another in the regional distribution of muscle mass. Madoqua kirkii (4-5 kg) concentrates muscle in the lumbar extensors and hindlimbs; large species such as Damaliscus doreas (50-60 kg) and Hippotragus niger (160-220 kg) distribute it more evenly between the lumbar and cervical regions and between the hindlimbs and forelimbs. Gazella dorcas (10-20 kg) is quantitatively intermediate in those characteristics. The redistribution of muscle mass with increasing size correlates with the loss of axial bending of the vertebral column: in small, hindlimb dominant, 'dorsomobile' species such as Madoqua sagittal mobility increases stride length through 'extended' suspension; in large 'dorsostable' species such as Damaliscus and Hippotragus the vertebral column resists bending, consequently abbreviating or omitting this non-contact phase of the gait cycle. Locomotor adaptation as it is reflected in size, shape, and musculoskeletal structure is the key to habitat choice, dietary specialization, social structure, and male agonistic behavior and, therefore, central to the fabric of behavioral ecology.