Yue Zhang

A novel multimodal solution is proposed to solve the problem of blind source separation (BSS) of moving sources. Since for moving sources the mixing filters are time varying, therefore, the unmixing filters should also be time varying and can be difficult to track in real time. In this solution the visual modality is utilized to facilitate the separation of moving sources. The movement of the sources is detected by a relatively simplistic 3-D tracker based on video cameras. The tracking process is based on particle filtering which provides robust tracking performance. Positions and velocities of the sources are obtained from the 3-D tracker and if the sources are moving, a beamforming algorithm is used to perform real time speech enhancement and provide separation of the sources. Experimental results show that by utilizing the visual modality, a good BSS performance for moving sources in a low reverberant environment can be achieved. © 2011, IGI Global.

As each type of satellite network has different link features, its data transmission must be designed based on its link features to improve the efficiency of data transferring. The transmission of navigation integrated services information (NISI) in a global navigation satellite system (GNSS) with inter-satellite links (ISLs) is studied by taking the real situation of inter-satellite communication links into account. An on-demand computing and buffering centralized route strategy is proposed based on dynamic grouping and the topology evolution law of the GNSS network within which the satellite nodes are operated in the manner of dynamic grouping. Dynamic grouping is based on satellites spatial relationships and the group role of the satellite node changes by turns due to its spatial relationships. The route strategy provides significant advantages of high efficiency, low complexity, and flexible configuration, by which the established GNSS can possess the features and capabilities of feasible deployment, efficient transmission, convenient management, structural invulnerability and flexible expansion.

With the fast development of the Internet, the size of Forwarding Information Base (FIB) maintained at backbone routers is experiencing an exponential growth, making the storage support and lookup process of FIBs a severe challenge. One effective way to address the challenge is FIB compression, and various solutions have been proposed in the literature. The main shortcoming of FIB compression is the overhead of updating the compressed FIB when routing update messages arrive. Only when the update time of FIB compression algorithms is small bounded can the probability of packet loss incurred by FIB compression operations during update be completely avoided. However, no prior FIB compression algorithm can achieve small bounded worst case update time, and hence a mature solution with complete avoidance of packet loss is still yet to be identified. To address this issue, we propose the Unite and Split (US) compression algorithm to enable fast update with controlled worst case update time. Further, we use the US algorithm to improve the performance of a number of classic software and hardware lookup algorithms. Simulation results show that the average update speed of the US algorithm is a little faster than that of the binary trie without any compression, while prior compression algorithms inevitably seriously degrade the update performance. After applying the US algorithm, the evaluated lookup algorithms exhibit significantly smaller on-chip memory consumption with little additional update overhead

In spite of the almost ideal variation of the radiative current of 1.3 mu m GaAsSb/GaAs-based lasers, the threshold current, J(th), is high due to non-radiative recombination accounting for 90% J(th) near room temperature. This also gives rise to low T-0 values similar to 60K close to room temperature, similar to that for InGaAsP/InP.

The highest power conversion efficiencies (PCEs) reported for perovskite solar cells (PSCs) with inverted planar structures are still inferior to those of PSCs with regular structures, mainly because of lower open-circuit voltages (Voc). Here we report a strategy to reduce nonradiative recombination for the inverted devices, based on a simple solution-processed secondary growth technique. This approach produces a wider bandgap top layer and a more n-type perovskite film, which mitigates nonradiative recombination, leading to an increase in Voc by up to 100 millivolts. We achieved a high Voc of 1.21 volts without sacrificing photocurrent, corresponding to a voltage deficit of 0.41 volts at a bandgap of 1.62 electron volts. This improvement led to a stabilized power output approaching 21% at the maximum power point.

A series of Ga(AsSb)/GaAs/(AlGa) As samples with varying GaAs spacer width are studied by electric-field modulated absorption (EA) and reflectance spectroscopy and modeled using a microscopic theory. The analysis of the Franz-Keldysh oscillations of GaAs capping layer and of the quantum-confined Stark shift of the lowest quantum well (QW) transitions shows the strong inhomogeneity of the built-in electric field indicating that the field modulation due to an external bias voltage differs significantly for the various regions of the structures. The calculations demonstrate that the line shape of the EA spectra of these samples is extremely sensitive to the value of the small conduction band offset between GaAs and Ga(AsSb) as well as to the magnitude of the internal electric field changes caused by the external voltage modulation in the QW region. The EA spectra of the entire series of samples are modeled by the microscopic theory. The good agreement between experiment and theory allows us to extract the strength of the modulation of the built-in electric field in the QW region and to show that the band alignment between GaAs and Ga(AsSb) is of type II with a conduction band offset of approximately 40 meV.

High-efficiency perovskite solar cells typically employ an organic–inorganic metal halide perovskite material as light absorber and charge transporter, sandwiched between a p-type electron-blocking organic hole-transporting layer and an n-type hole-blocking electron collection titania compact layer. Some device configurations also include a thin mesoporous layer of TiO2 or Al2O3 which is infiltrated and capped with the perovskite absorber. Herein, we demonstrate that it is possible to fabricate planar and mesoporous perovskite solar cells devoid of an electron selective hole-blocking titania compact layer, which momentarily exhibit power conversion efficiencies (PCEs) of over 13%. This performance is however not sustained and is related to the previously observed anomalous hysteresis in perovskite solar cells. The “compact layer-free” meso-superstructured perovskite devices yield a stabilised PCE of only 2.7% while the compact layer-free planar heterojunction devices display no measurable steady state power output when devoid of an electron selective contact. In contrast, devices including the titania compact layer exhibit stabilised efficiency close to that derived from the current voltage measurements. We propose that under forward bias the perovskite diode becomes polarised, providing a beneficial field, allowing accumulation of positive and negative space charge near the contacts, which enables more efficient charge extraction. This provides the required built-in potential and selective charge extraction at each contact to temporarily enable efficient operation of the perovskite solar cells even in the absence of charge selective n- and p-type contact layers. The polarisation of the material is consistent with long range migration and accumulation of ionic species within the perovskite to the regions near the contacts. When the external field is reduced under working conditions, the ions can slowly diffuse away from the contacts redistributing throughout the film, reducing the field asymmetry and the effectiveness of the operation of the solar cells. We note that in light of recent publications showing high efficiency in devices devoid of charge selective contacts, this work reaffirms the absolute necessity to measure and report the stabilised power output under load when characterizing perovskite solar cells.

Isochronous mass spectrometry was applied to measure isomeric yield ratios of fragmentation reaction prod- ucts. This approach is complementary to conventional g -ray spectroscopy in particular for measuring yield ratios for long-lived isomeric states. Isomeric yield ratios for the high-spin I = 19=2¯h states in the mirror nuclei 53Fe and 53Co are measured to study angular momentum population following the projectile fragmentation of 78Kr at energies of 480 A MeV on a beryllium target. The 19/2 state isomeric ratios of 53Fe produced from different projectiles in literature have also been extracted as a function of mass number difference between projectile and fragment (mass loss). The results are compared to ABRABLA07 model calculations. The isomeric ratios of 53Fe produced using different projectiles suggest that the theory underestimates not only the previously reported dependence on the spin but also the dependence on the mass loss.

We propose a multi-view framework for joint object detection and labelling based on pairs of images. The proposed framework extends the single-view Mask R-CNN approach to multiple views without need for additional training. Dedicated components are embedded into the framework to match objects across views by enforcing epipolar constraints, appearance feature similarity and class coherence. The multi-view extension enables the proposed framework to detect objects which would otherwise be mis-detected in a classical Mask R-CNN approach, and achieves coherent object labelling across views. By avoiding the need for additional training, the approach effectively overcomes the current shortage of multi-view datasets. The proposed framework achieves high quality results on a range of complex scenes, being able to output class, bounding box, mask and an additional label enforcing coherence across views. In the evaluation, we show qualitative and quantitative results on several challenging outd oor multi-view datasets and perform a comprehensive comparison to verify the advantages of the proposed method