ngoc duy Nguyen

ABSTRACT Person Re-Identification refers to recognizing people across cameras with non-overlapping capture areas. To recognize people, their images must be represented by feature vectors for matching. Recent state-of-the-art approaches employ semantic features, also known as attributes (e.g. wearing-bags, jeans, skirt), for presentation. However, such presentations are sensitive to attribute detection results which can be irrelevant due to noise. In this paper, we propose an approach to exploit relationships between attributes for refining attribute detection results. Experimental results on benchmark datasets (VIPeR and PRID) demonstrate the effectiveness of our proposed approach.

ABSTRACT Person re-identification is the problem of matching pedestrian images observed by different cameras in non-overlapping regions. Semantic features, also called attributes, have demonstrated to produce state-of-the-art performances in this problem. In existing works, attributes are detected independently to each other. In this paper, we propose using relationships between attributes to refine the attribute detection result. Experimental results on two datasets VIPeR and PRID prove the effectiveness on performances when our method is applied.

ABSTRACT A key issue in the commercialization of dye-sensitized solar cells is to maintain high efficiency and long lifetime. As reported in the literature, dye-sensitized solar cells are stable under visible light soaking but thermal stress and UV exposure lead to efficiency degradation. However, all the stability studies published so far have been performed on cells whose TiO2 electrodes were prepared by tape casting or screen printing of nanoparticle pastes/inks. The present study concerns cells based on highly porous templated TiO2 electrodes, whose larger surface area could enhance the negative effects of thermal stress, light soaking and UV exposure. The long-term stability of these cells is compared with a classical nanoparticle-based cell using current-voltage measurements (I-V curves) and electrochemical impedance spectroscopy. Due to their higher active interface, templated cells are more sensitive than nanoparticle cells to UV illumination, although this can be easily solved in both cases by the use of a UV filter. The templated cells are as stable as the nanoparticle cells under visible light soaking (UV filtered). However, we showed that templated cells are more stable under thermal stress. Moreover, as evidenced by electrochemical impedance spectroscopy, templated cells show lower transfer resistance, as well as lower recombination resistance compared to nanoparticle cells. The crystallite connectivity promoted by the templating route seems to favor the electron transfers inside the porous layer. Using templated films in dye-sensitized solar cells is therefore really promising because higher conversion efficiencies are reached without promoting cell degradation.

ABSTRACT Evolved Multimedia Broadcast Multicast Service (eMBMS) is a point-to-multipoint content delivery solution designed for LTE/LTE-A. It distributes efficiently the broadcast and multicast service to a massive number of mobile devices located in a given geographical area. eMBMS can be used to boost the network's capability for providing high-quality multimedia service in high user-density area such a stadium during an entertainment or sport event. The aim of this paper is to validate and evaluate the performance of eMBMS following the 3GPP standard (release 10) implemented in the context of the OpenAirInterface SDR platforms. The difference between the OpenAirInterface in-lab system validation platform with respect to existing simulation/emulation tools such as OPNET, matlab, NS-2/3 is that firstly it is built with a high level of realism supporting real-time operation (based on RTAI and RT-Preempt); secondly it is part of the integration and validation chain for a real-time RF experimentation. The results show that eMBMS performance in the OpenAirInterface satisfies the requirement of 3GPP standard in terms of BLER. Furthermore, the theoretical user throughput proposed by 3GPP have been validated through experimentation.

ABSTRACT Evolved Multimedia Broadcast Multicast Service (eMBMS) has recently attracted a great attention from the telecommunication industry. All big companies in the field have invested and intended to deploy eMBMS as the broadcast solution in response to the immense demand in multimedia traffic. The eMBMS, also known as LTE broadcast, can provide high quality live video streaming services for a large number of users simultaneously in an open area such as a stadium or arena during an entertainment or sport event. Nowadays, in a high mobility environment, maintaining the service reception for moving users becomes a crucial task for mobile operators and service providers. Unfortunately, the support from LTE standard for eMBMS service continuity is very limited at the moment and in many cases, the users cannot continue to receive their desired services. To overcome the limitation in the standard, this paper will present a novel method to ensure the service continuity as well as to reduce the service interruption time during handover period for eMBMS users in the mobility context.

We investigated the growth of in-situ n-type doped epitaxial Si layers with arsenic and phosphorus by means of low-temperature chemical vapor deposition using trisilane as Si-precursor. Indeed, in order to prevent the alteration of the characteristics of the devices which are already present on the wafer, an epitaxy process at low temperature is highly desired for applications such as BiCMOS. In this work, the varying parameters are the deposition temperature, the Si-precursor mass flow and the dopant gas flow. As a result, a process for the deposition of heavily doped epilayers was demonstrated at 600 °C with high deposition rate, which is important for maintaining high throughput and low process cost. We showed that using trisilane as a Si-precursor resulted in a much more linear n-type doping behavior than using dichlorosilane. Therefore it allowed an easier process control and a wider dynamic doping range. Our process is an interesting route for the epitaxy of a low-resistance emitter layer for bipolar transistor application.

The authors demonstrated that the combination of VPD and LA enables the fabrication of high quality, defect-free USJs with abrupt dopant profile. The results for PMOS with B-VPD are very promising for the 32 nm and the 22 nm technology nodes. In the case of NMOS, As-VPD and LA enable the fabrication of an USJ but the electrical deactivation of a large part of the in-diffused dopants is responsible for the high sheet resistance values.

We evaluated the combination of vapor phase doping and sub-melt laser anneal as a novel doping strategy for the fabrication of source and drain extension junctions in sub-32 nm CMOS technology, aiming at both planar and non-planar device applications. High quality ultra shallow junctions with abrupt profiles in Si substrates were demonstrated on 300 mm Si substrates. The excellent results obtained for the sheet resistance and the junction depth with boron allowed us to fulfill the requirements for the 32 nm as well as for the 22 nm technology nodes in the PMOS case by choosing appropriate laser anneal conditions. For instance, using 3 laser scans at 1300 °C, we measured an active dopant concentration of about 2.1 × 1020 cm− 3 and a junction depth of 12 nm. With arsenic for NMOS, ultra shallow junctions were achieved as well. However, as also seen for other junction fabrication schemes, low dopant activation level and active dose (in the range of 1–4 × 1013 cm− 2) were observed although dopant concentration versus depth profiles indicate that the dopant atoms were properly driven into the substrate during the anneal step. The electrical deactivation of a large part of the in-diffused dopants was responsible for the high sheet resistance values.

ABSTRACT The ITRS Roadmap highlights the electrical characterization of the source and drain extension regions as a key challenge for future complimentary-metal-oxide-semiconductor technology. Presently, an accurate determination of the depth of ultrashallow junctions can routinely only be performed by time-consuming and destructive techniques such as secondary ion mass spectrometry SIMS. In this work, the authors propose to use the fast and nondestructive photomodulated optical reflectance PMOR technique , as implemented in the Therma-Probe ® TP dopant metrology system, for these purposes. PMOR is a pump-probe technique based on the measurement of the pump-induced modulated change in probe reflectance, i.e., the so-called photo modulated reflectance. In this article, the authors demonstrate that the absolute junction depths of boxlike active dopant structures can be extracted in a very simple and straightforward way from the TP offset curves, which represent the behavior of the modulated reflectance as a function of the pump-probe beam spacing. Although the procedure is based on the insights into the physical behavior of the offset curves, no modeling is involved in the actual extraction process itself. The extracted junction depths are in good correlation with the corresponding junction depths as measured by means of SIMS. The technique has a subnanometer depth sensitivity for depths ranging from 10 to 35 nm with the present Therma-Probe ® 630XP system. The extension of the proposed procedure to the general ultrashallow profiles is also explored and discussed. © 2010 American Vacuum Society.

In this paper, a numerical simulation method was used, based on the solution of the basic semiconductor equations. This gives an access to microscopic and macroscopic properties of the structure, and thereon, to an understanding of the electrical properties of Ge1-xSnx/Ge by linking quantities such as admittance spectra to microscopic variations in the structure. A 200 nm thick p-doped Ge1-xSnx layer (x≈0.05) on top of a 100 μm por n-Ge substrate was modeled, including a Shockley Read Hall trap with energy near the valence band.

In wireless sensor networks, density control is a promising approach that has a considerable impact on extending the network lifetime. To optimize the energy utilization, the sensors in active mode adjust their sensing ranges to fulfill two conflicting objectives, minimizing the overlapping sensing area of the sensors and at the same time maintaining a high degree of coverage. In this

ABSTRACT I-V characteristics of pGeSn/nGe diodes have been measured and show very interesting properties. Simulations of the same structure are able to reproduce most of the observed behavior and point to the predominant influence of parameters such as the band gap energy of the GeSn layer. C-V characteristics showing little frequency dependence have also been measured, and their analysis for the determination of the carrier concentration is confirmed by simulations. More investigations of the effect of temperature, of other observed features in the C-V characteristics and of other defects at the interface or in the bulk of either layers, are still required in order to explain some of the observed behaviors, notably the reverse saturation current.

The uneven deployment of wireless sensor networks and the breakdown of sensor nodes often lead to the presence of holes in their topology. Two important factors considered in designing a geographic routing algorithm are overhead and success rate of packet routing. Therefore, in this paper, we present a novel strategy to aid a routing algorithm improve its performance by increasing

The coverage problem is a fundamental issue in wireless sensor networks. It has attracted considerable attention recently. Most node scheduling patterns utilize the adjustable range of sensors to minimize the sensing energy consumption, and thus extend the network lifetime. However, a large source of the consumption of the sensor communication energy is not strictly taken into account. In this paper,

The coverage problem is a fundamental issue in wireless sensor networks. It has attracted considerable attention recently. Most node scheduling patterns utilize the adjustable range of sensors to minimize the sensing energy consumption, and thus extend the network lifetime. However, a large source of the consumption of the sensor communication energy is not strictly taken into account. In this paper, we introduce two energy-efficient patterns that are used to minimize the communication energy consumption of a sensor network, and simultaneously, maintain a high degree of coverage. Moreover, the proposed patterns have a structure that is easy to design and apply to practical applications. Calculations and extensive simulation are conducted to evaluate the efficiency of the new patterns compared to existing ones in terms of various performance metrics.

In this work we optimized the Ge-spiked monoemitter for the SiGe:C heterojunction bipolar transistor by using low-temperature trisilane based chemical vapor deposition to meet the requirements of high growth rate and high electrically-active doping levels of arsenic. The resultant devices show improvement of open-base breakdown voltage and no degradation of cutoff frequency with incorporation of a Ge spike in the monoemitter.

In this paper we present results from an InGaAs/InP implant free quantum well device integrated fully in a Si CMOS processing line. The virtual InP substrates are generated using a Si template which is prepared by standard STI processing. The Si in the STI trenches is etched and a Ge seed layer grown.