Altres
http://hdl.handle.net/2117/15797
Tue, 21 Nov 2017 06:57:52 GMT
20171121T06:57:52Z

Admission control for multitenant radio access networks
http://hdl.handle.net/2117/110958
Admission control for multitenant radio access networks
Pérez Romero, Jordi; Sallent Roig, José Oriol; Ferrús Ferré, Ramón Antonio; Agustí Comes, Ramon
The sharing of Radio Access Networks is gaining momentum in small cell scenarios, due to the associated reduction in capital and operational costs. In this scenario, the split of radio resources among tenants sharing the network becomes a fundamental problem to provide each one with its required capacity. In this respect, this paper proposes a multitenant admission control scheme for performing this split with the target of ensuring an efficient use of the radio resources by exploiting traffic multiplexing principles at both intracell and multicell level in order to cope with heterogeneities in the spatial traffic distribution of the different tenants. A simulationbased analysis is presented to assess the flexibility of the proposed approach under different traffic mixes of the different tenants in each cell. Substantial bit rate improvements (e.g. up to 106%) and blocking probability reductions are obtained with respect to a baseline scheme.
Mon, 20 Nov 2017 19:37:11 GMT
http://hdl.handle.net/2117/110958
20171120T19:37:11Z
Pérez Romero, Jordi
Sallent Roig, José Oriol
Ferrús Ferré, Ramón Antonio
Agustí Comes, Ramon
The sharing of Radio Access Networks is gaining momentum in small cell scenarios, due to the associated reduction in capital and operational costs. In this scenario, the split of radio resources among tenants sharing the network becomes a fundamental problem to provide each one with its required capacity. In this respect, this paper proposes a multitenant admission control scheme for performing this split with the target of ensuring an efficient use of the radio resources by exploiting traffic multiplexing principles at both intracell and multicell level in order to cope with heterogeneities in the spatial traffic distribution of the different tenants. A simulationbased analysis is presented to assess the flexibility of the proposed approach under different traffic mixes of the different tenants in each cell. Substantial bit rate improvements (e.g. up to 106%) and blocking probability reductions are obtained with respect to a baseline scheme.

Joint optimization of path selection and link scheduling for millimeter wave transport networks
http://hdl.handle.net/2117/110952
Joint optimization of path selection and link scheduling for millimeter wave transport networks
Huerfano, Diego; Seyfettin Demirkol, Ilker; Legg, Peter
In future mobile networks, the wireless transport networks are expected to carry traffic flows with different throughput and delay requirements due to the introduction of CloudRAN (CRAN) and different functional splits that can be used, e.g., as defined by Next Generation Fronthaul Interface (NGFI). A promising wireless technology to support the high throughput requirements of CRAN splits is Millimeter Wave (mmWave) band technologies standardized by IEEE 802.11ad amendment. Our target here is to derive the mathematical formulation of the path selection and link scheduling problem for mmWave transport networks, where the backhaul and fronthaul flows will coexist, by defining the constraints brought by different functional splits and the IEEE 802.11ad standard. We present two objective functions that can be used for this problem: load balancing and minimization of the use of air time. We implemented the derived formulations in an MixedInteger Linear Programming (MILP) solver and evaluated realistic scenarios of wireless fronthaul/backhaul networks assessing the splits defined by NGFI for LTE.
Mon, 20 Nov 2017 19:14:41 GMT
http://hdl.handle.net/2117/110952
20171120T19:14:41Z
Huerfano, Diego
Seyfettin Demirkol, Ilker
Legg, Peter
In future mobile networks, the wireless transport networks are expected to carry traffic flows with different throughput and delay requirements due to the introduction of CloudRAN (CRAN) and different functional splits that can be used, e.g., as defined by Next Generation Fronthaul Interface (NGFI). A promising wireless technology to support the high throughput requirements of CRAN splits is Millimeter Wave (mmWave) band technologies standardized by IEEE 802.11ad amendment. Our target here is to derive the mathematical formulation of the path selection and link scheduling problem for mmWave transport networks, where the backhaul and fronthaul flows will coexist, by defining the constraints brought by different functional splits and the IEEE 802.11ad standard. We present two objective functions that can be used for this problem: load balancing and minimization of the use of air time. We implemented the derived formulations in an MixedInteger Linear Programming (MILP) solver and evaluated realistic scenarios of wireless fronthaul/backhaul networks assessing the splits defined by NGFI for LTE.

The appraisal of machine learning techniques for tourism demand forecasting
http://hdl.handle.net/2117/110947
The appraisal of machine learning techniques for tourism demand forecasting
Claveria, Oscar; Monte Moreno, Enrique; Torra Porras, Salvador
Machine learning (ML) methods are being increasingly used with forecasting purposes. This study assesses the predictive performance of several ML models in a multipleinput multipleoutput (MIMO) setting that allows incorporating the crosscorrelations between the inputs. We compare the forecast accuracy of a Gaussian process regression (GPR) model to that of different neural network architectures in a multistepahead time series prediction experiment. We find that the radial basis function (RBF) network outperforms the GPR model, especially for longterm forecast horizons. As the memory of the models increases, the forecasting performance of the GPR improves, suggesting the convenience of designing a model selection criteria in order to estimate the optimal number of lags used for concatenation.
Mon, 20 Nov 2017 18:53:20 GMT
http://hdl.handle.net/2117/110947
20171120T18:53:20Z
Claveria, Oscar
Monte Moreno, Enrique
Torra Porras, Salvador
Machine learning (ML) methods are being increasingly used with forecasting purposes. This study assesses the predictive performance of several ML models in a multipleinput multipleoutput (MIMO) setting that allows incorporating the crosscorrelations between the inputs. We compare the forecast accuracy of a Gaussian process regression (GPR) model to that of different neural network architectures in a multistepahead time series prediction experiment. We find that the radial basis function (RBF) network outperforms the GPR model, especially for longterm forecast horizons. As the memory of the models increases, the forecasting performance of the GPR improves, suggesting the convenience of designing a model selection criteria in order to estimate the optimal number of lags used for concatenation.

Computing for democracy: the Asociación de Técnicos de Informática and the professionalization of computing in Spain
http://hdl.handle.net/2117/110945
Computing for democracy: the Asociación de Técnicos de Informática and the professionalization of computing in Spain
Fornés de Juan, Jordi; Herran, Nestor; Duque, Luís
The professionalization of computing in Spain had a key player in the Association of Information Technology Technicians (ATI), that fought for better working conditions, promotion of academic programs in computer science, and lobbying for the regulation of computing.
© 2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Mon, 20 Nov 2017 18:40:15 GMT
http://hdl.handle.net/2117/110945
20171120T18:40:15Z
Fornés de Juan, Jordi
Herran, Nestor
Duque, Luís
The professionalization of computing in Spain had a key player in the Association of Information Technology Technicians (ATI), that fought for better working conditions, promotion of academic programs in computer science, and lobbying for the regulation of computing.

On average real sliding dynamics in linear systems
http://hdl.handle.net/2117/110940
On average real sliding dynamics in linear systems
Olm Miras, Josep Maria; Biel Solé, Domingo; Repecho del Corral, Víctor; Shtessel, Yuri B.
It is well known that in implementations of sliding mode controllers using hysteresis comparators, when the hysteresis band amplitude tends to zero the real dynamics tends to the ideal sliding dynamics. However, in real systems physical limitations do not allow to effectively lower this value at will, and a steady state error is likely to appear. In this paper we relate this error with a non zero average value of the switching function in each switching period: it is shown that, in linear systems, when the controller has a constant switching frequency and the switching function is periodic, the average value of the difference between real and ideal steady state dynamics is proportional to the average value of the switching function. Hence, when this average value is non zero an average steady state error appears, while a zero average value for the switching function entails no average steady state error. The proof is carried out using a regular form approach, and the result is exemplified in a buck converter. Simulation results show that when the switching function is periodic and shows a piecewise linear behavior within the hysteresis band, thus guaranteeing zero average value, the average state error disappears. In turn, when this piecewise linear character is lost and the switching function has non zero mean value, an average steady state error arises.
Mon, 20 Nov 2017 16:33:09 GMT
http://hdl.handle.net/2117/110940
20171120T16:33:09Z
Olm Miras, Josep Maria
Biel Solé, Domingo
Repecho del Corral, Víctor
Shtessel, Yuri B.
It is well known that in implementations of sliding mode controllers using hysteresis comparators, when the hysteresis band amplitude tends to zero the real dynamics tends to the ideal sliding dynamics. However, in real systems physical limitations do not allow to effectively lower this value at will, and a steady state error is likely to appear. In this paper we relate this error with a non zero average value of the switching function in each switching period: it is shown that, in linear systems, when the controller has a constant switching frequency and the switching function is periodic, the average value of the difference between real and ideal steady state dynamics is proportional to the average value of the switching function. Hence, when this average value is non zero an average steady state error appears, while a zero average value for the switching function entails no average steady state error. The proof is carried out using a regular form approach, and the result is exemplified in a buck converter. Simulation results show that when the switching function is periodic and shows a piecewise linear behavior within the hysteresis band, thus guaranteeing zero average value, the average state error disappears. In turn, when this piecewise linear character is lost and the switching function has non zero mean value, an average steady state error arises.

Word ordering and document adjacency for large loop closure detection in 2D laser maps
http://hdl.handle.net/2117/110932
Word ordering and document adjacency for large loop closure detection in 2D laser maps
Deray, Jeremie; Solà Ortega, Joan; AndradeCetto, Juan
We address in this paper the problem of loop closure detection for laserbased simultaneous localization and mapping (SLAM) of very large areas. Consistent with the state of the art, the map is encoded as a graph of poses, and to cope with very large mapping capabilities, loop closures are asserted by comparing the features extracted from a query laser scan against a previously acquired corpus of scan features using a bagofwords (BoW) scheme. Two contributions are here presented. First, to benefit from the graph topology, feature frequency scores in the BoW are computed not only for each individual scan but also from neighboring scans in the SLAM graph. This has the effect of enforcing neighbor relational information during document matching. Secondly, a weak geometric check that takes into account feature ordering and occlusions is introduced that substantially improves loop closure detection performance. The two contributions are evaluated both separately and jointly on four common SLAM datasets, and are shown to improve the stateoftheart performance both in terms of precision and recall in most of the cases. Moreover, our current implementation is designed to work at nearly frame rate, allowing loop closure query resolution at nearly 22 Hz for the best case scenario and 2 Hz for the worst case scenario.
© 20xx IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works
Mon, 20 Nov 2017 15:39:15 GMT
http://hdl.handle.net/2117/110932
20171120T15:39:15Z
Deray, Jeremie
Solà Ortega, Joan
AndradeCetto, Juan
We address in this paper the problem of loop closure detection for laserbased simultaneous localization and mapping (SLAM) of very large areas. Consistent with the state of the art, the map is encoded as a graph of poses, and to cope with very large mapping capabilities, loop closures are asserted by comparing the features extracted from a query laser scan against a previously acquired corpus of scan features using a bagofwords (BoW) scheme. Two contributions are here presented. First, to benefit from the graph topology, feature frequency scores in the BoW are computed not only for each individual scan but also from neighboring scans in the SLAM graph. This has the effect of enforcing neighbor relational information during document matching. Secondly, a weak geometric check that takes into account feature ordering and occlusions is introduced that substantially improves loop closure detection performance. The two contributions are evaluated both separately and jointly on four common SLAM datasets, and are shown to improve the stateoftheart performance both in terms of precision and recall in most of the cases. Moreover, our current implementation is designed to work at nearly frame rate, allowing loop closure query resolution at nearly 22 Hz for the best case scenario and 2 Hz for the worst case scenario.

Kinodynamic planning on constraint manifolds
http://hdl.handle.net/2117/110931
Kinodynamic planning on constraint manifolds
Bordalba Llaberia, Ricard; Porta Pleite, Josep Maria; Ros Giralt, Lluís
This report presents a motion planner for systems subject to kinematic and dynamic constraints. The former appear when kinematic loops are present in the system, such as in parallel manipulators, in robots that cooperate to achieve a given task, or in situations involving contacts with the environment. The latter are necessary to obtain realistic trajectories, taking into account the forces acting on the system. The kinematic constraints make the state space become an implicitlydefined manifold, which complicates the application of common motion planning techniques. To address this issue, the planner constructs an atlas of the state space manifold incrementally, and uses this atlas both to generate random states and to dynamically simulate the steering of the system towards such states. The resulting tools are then exploited to construct a rapidlyexploring random tree (RRT) over the state space. To the best of our knowledge, this is the first randomized kinodynamic planner for implicitlydefined state spaces. The test cases presented validate the approach in significantlycomplex systems.
Mon, 20 Nov 2017 15:33:22 GMT
http://hdl.handle.net/2117/110931
20171120T15:33:22Z
Bordalba Llaberia, Ricard
Porta Pleite, Josep Maria
Ros Giralt, Lluís
This report presents a motion planner for systems subject to kinematic and dynamic constraints. The former appear when kinematic loops are present in the system, such as in parallel manipulators, in robots that cooperate to achieve a given task, or in situations involving contacts with the environment. The latter are necessary to obtain realistic trajectories, taking into account the forces acting on the system. The kinematic constraints make the state space become an implicitlydefined manifold, which complicates the application of common motion planning techniques. To address this issue, the planner constructs an atlas of the state space manifold incrementally, and uses this atlas both to generate random states and to dynamically simulate the steering of the system towards such states. The resulting tools are then exploited to construct a rapidlyexploring random tree (RRT) over the state space. To the best of our knowledge, this is the first randomized kinodynamic planner for implicitlydefined state spaces. The test cases presented validate the approach in significantlycomplex systems.

A distributed predictive control approach for periodic flowbased networks: application to drinking water systems
http://hdl.handle.net/2117/110930
A distributed predictive control approach for periodic flowbased networks: application to drinking water systems
Grosso Perez, Juan Manuel; OcampoMartínez, Carlos; Puig Cayuela, Vicenç
This paper proposes a distributed model predictive control approach designed to work in a cooperative manner for controlling flowbased networks showing periodic behaviours. Under this distributed approach, local controllers cooperate in order to enhance the performance of the whole flow network avoiding the use of a coordination layer. Alternatively, controllers use both the monolithic model of the network and the given global cost function to optimise the control inputs of the local controllers but taking into account the effect of their decisions over the remainder subsystems conforming the entire network. In this sense, a global (alltoall) communication strategy is considered. Although the Pareto optimality cannot be reached due to the existence of nonsparse coupling constraints, the asymptotic convergence to a Nash equilibrium is guaranteed. The resultant strategy is tested and its effectiveness is shown when applied to a largescale complex flowbased network: the Barcelona drinking water supply system.
Mon, 20 Nov 2017 15:30:05 GMT
http://hdl.handle.net/2117/110930
20171120T15:30:05Z
Grosso Perez, Juan Manuel
OcampoMartínez, Carlos
Puig Cayuela, Vicenç
This paper proposes a distributed model predictive control approach designed to work in a cooperative manner for controlling flowbased networks showing periodic behaviours. Under this distributed approach, local controllers cooperate in order to enhance the performance of the whole flow network avoiding the use of a coordination layer. Alternatively, controllers use both the monolithic model of the network and the given global cost function to optimise the control inputs of the local controllers but taking into account the effect of their decisions over the remainder subsystems conforming the entire network. In this sense, a global (alltoall) communication strategy is considered. Although the Pareto optimality cannot be reached due to the existence of nonsparse coupling constraints, the asymptotic convergence to a Nash equilibrium is guaranteed. The resultant strategy is tested and its effectiveness is shown when applied to a largescale complex flowbased network: the Barcelona drinking water supply system.

ROS wrapper for realtime multiperson pose estimation with a single camera
http://hdl.handle.net/2117/110929
ROS wrapper for realtime multiperson pose estimation with a single camera
Arduengo García, Miguel; Jorgensen, Steven Jens; Hambuchen, Kimberly; Sentis, Luis; MorenoNoguer, Francesc; Alenyà Ribas, Guillem
For robots to be deployable in human occupied environments, the robots must have humanawareness and generate humanaware behaviors and policies. OpenPose is a library for realtime multiperson keypoint detection. We have considered the implementation of a ROS package that would allow the estimation of 2d pose from simple RGB images, for which we have introduced a ROS wrapper that automatically recovers the pose of several people from a single camera using OpenPose. Additionally, a ROS node to obtain 3d pose estimation from the initial 2d pose estimation when a depth image is synchronized with the RGB image (RGBD image, such as with a Kinect camera) has been developed. This aim is attained projecting the 2d pose estimation onto the pointcloud of the depth image.
Mon, 20 Nov 2017 15:26:26 GMT
http://hdl.handle.net/2117/110929
20171120T15:26:26Z
Arduengo García, Miguel
Jorgensen, Steven Jens
Hambuchen, Kimberly
Sentis, Luis
MorenoNoguer, Francesc
Alenyà Ribas, Guillem
For robots to be deployable in human occupied environments, the robots must have humanawareness and generate humanaware behaviors and policies. OpenPose is a library for realtime multiperson keypoint detection. We have considered the implementation of a ROS package that would allow the estimation of 2d pose from simple RGB images, for which we have introduced a ROS wrapper that automatically recovers the pose of several people from a single camera using OpenPose. Additionally, a ROS node to obtain 3d pose estimation from the initial 2d pose estimation when a depth image is synchronized with the RGB image (RGBD image, such as with a Kinect camera) has been developed. This aim is attained projecting the 2d pose estimation onto the pointcloud of the depth image.

The use of a binary composite endpoint and sample size requirement: influence of endpoints overlap
http://hdl.handle.net/2117/110928
The use of a binary composite endpoint and sample size requirement: influence of endpoints overlap
Ramon Marsal, Josep; Ferreira González, Ignacio; Bertrán, Sandra; Ribera, Aida; PermanyerMiralda, Gaietà; Garcia Dorado, Antonio David; Gómez Melis, Guadalupe
Although composite endpoints (CE) are common in clinical trials, the impact of the relationship between the components of a binary CE on the sample size requirement (SSR) has not been addressed. We performed a computational study considering 2 treatments and a CE with 2 components: the relevant endpoint (RE) and the additional endpoint (AE). We assessed the strength of the components’ interrelation by the degree of relative overlap between them, which was stratified into 5 groups. Within each stratum, SSR was computed for multiple scenarios by varying the events proportion and the effect of the therapy. A lower SSR using CE was defined as the best scenario for using the CE. In 25 of 66 scenarios the degree of relative overlap determined the benefit of using CE instead of the RE. Adding an AE with greater effect than the RE leads to lower SSR using the CE regardless of the AE proportion and the relative overlap. The influence of overlapping decreases when the effect on RE increases. Adding an AE with lower effect than the RE constitutes the most uncertain situation. In summary, the interrelationship between CE components, assessed by the relative overlap, can help to define the SSR in specific situations and it should be considered for SSR computation.
Mon, 20 Nov 2017 15:21:50 GMT
http://hdl.handle.net/2117/110928
20171120T15:21:50Z
Ramon Marsal, Josep
Ferreira González, Ignacio
Bertrán, Sandra
Ribera, Aida
PermanyerMiralda, Gaietà
Garcia Dorado, Antonio David
Gómez Melis, Guadalupe
Although composite endpoints (CE) are common in clinical trials, the impact of the relationship between the components of a binary CE on the sample size requirement (SSR) has not been addressed. We performed a computational study considering 2 treatments and a CE with 2 components: the relevant endpoint (RE) and the additional endpoint (AE). We assessed the strength of the components’ interrelation by the degree of relative overlap between them, which was stratified into 5 groups. Within each stratum, SSR was computed for multiple scenarios by varying the events proportion and the effect of the therapy. A lower SSR using CE was defined as the best scenario for using the CE. In 25 of 66 scenarios the degree of relative overlap determined the benefit of using CE instead of the RE. Adding an AE with greater effect than the RE leads to lower SSR using the CE regardless of the AE proportion and the relative overlap. The influence of overlapping decreases when the effect on RE increases. Adding an AE with lower effect than the RE constitutes the most uncertain situation. In summary, the interrelationship between CE components, assessed by the relative overlap, can help to define the SSR in specific situations and it should be considered for SSR computation.