ognition in simple scenes, the difference between classes of basketball actions is slight, and the backgrounds in the video are very similar. Therefore, it is not easy to recognize the basketball actions directly based on short-term temporal information or the scene information in the video. A Global Context-Aware Network (GCA-Net) for basketball action recognition is proposed to address this problem in this paper. It contains a Multi-Time Scale Aggregation (MTSA) module and a Spatial-Channel Interaction (SCI) module to process multiple types of information on feature layers. The MTSA module uses a temporal pyramid to get contextual links in the temporal dimension through one-dimensional convolution with different dilation rates. The SCI module enhances the feature representation to obtain more prosperous category attributes and spatial information by interacting with information across dimensions. We conducted experiments on the basketball action recognition dataset SpaceJam, and the results show that GCA-Net can effectively classify basketball actions. The average recognition accuracy of ten types of basketball actions in the dataset is 91.54%, which is an improvement compared with the current mainstream methods.
In order to ensure the safe driving, autonomous vehicles under dynamic environments must accurately identify important targets that may change or move, especially pedestrians. Because the cameras are vulnerable to weather, light and other environmental factors, the real-time and accuracy of target detection and recognition may be poor. In this paper, we propose a method for pedestrian target recognition based on Lidar. It consists of three parts: ground point cloud removal, target object segmentation and pedestrian target recognition. Firstly, due to the huge amount of ground point cloud data, this paper uses a range image based on angle threshold algorithm to rapidly remove ground point cloud. With the mapping relationship between the disordered 3D point cloud and the ordered range image, the ground point cloud is marked and removed by the angle threshold on the range image rather than on the 3D point cloud. The ground point cloud removal is convenient for subsequent target object segmentation and recognition. Secondly, because the traditional clustering segmentation algorithm can not meet the real-time requirement of point cloud segmentation, we use improved Eps parameter DBSCAN algorithm combined with range image to segment point cloud target in this paper. For the problem of non-uniformity of 3D point cloud data density, we improve the setting of Eps parameter. For the undersegmentation between adjacent targets, we consider both Euclidean space distance and the angular distance which can distinguish the adjacent targets well in the space. Finally, considering the sparseness and disorder of the 3D point cloud and the lack of local feature extraction of the traditional PointNet network, a multi-scale feature fusion PointNet network is proposed, which combines the multi-scale local features and global feature of point cloud. The network also uses the spatial information and reflection intensity of Lidar point cloud to complete the pedestrian recognition. The experimental result shows that our proposed method can rapidly remove the ground point cloud and works well in segmentation between adjacent targets. The multi-scale feature fusion PointNet network performances well on 3D point cloud, its Area Under the Curve of ROC value reaches 0.92.
Hybrid waveband/wavelength switching in intelligent optical networks is gaining more and more academic attention. It
is very challenging to develop efficient algorithms to efficiently use waveband switching capability. In this paper, we
propose a novel cross-layer routing algorithm, waveband layered graph routing algorithm (WBLGR), in waveband
switching-enabled optical networks. Through extensive simulation WBLGR algorithm can significantly improve the
performance in terms of reduced call blocking probability.
Giving a new method to tuning a fiber Bragg grating into a chirped grating using a gradient curvature beam. Stick a FBG to an "S" shape beam with one end free and another give a pull. The curvature of the beam would change and the FBG get gradient strain. Thus we get the chirped grating while the central wavelength doesn't change. The design method and the theoretical analysis are given. The maximum chirped bandwidth in our experiment is 5.10nm and the central wavelength does not change.
A novel waveband switching node architecture is proposed using tunable optic filters for arbitrary waveband add/drop. The simulation results show this new architecture achieves lower blocking probability and reduce the scale of ports at switch fabric.
In OBS, JET (Just-Enough-Time) is the classical wavelength reservation scheme. But there is a phenomenon that the burst priority decreasing hop-by-hop in multi-hop networks that will waste the bandwidth that was used in the upstream. Based on the HPI (Hop-by-hop Priority Increasing) proposed in the former research, this paper will do an unprecedented simulation in 4×4 meshed topology, which is closer to the real network environment with the help of a NS2-based OBSN simulation platform constructed by ourselves. By contrasting, the drop probability and throughput on one of the longest end-to-end path lengths in the whole networks, it shows that the HPI scheme can improve the utilance of bandwidth better.
Recently, performance evaluation of TCP over Optical Burst-switched Network (OBSN) has drawn more and more attention, but previous work mainly focused on single TCP flow or static traffic model. In this paper, performances of different TCP implementations deployed in OBSN under dynamic competing traffic are investigated.
Waveband switching routing is a new problem for research community. It is has some different with the routing of wavelength assign problem. In this paper, we propose a new ideal for agent based Waveband switching network construction. Primary simulation shows that in this way complexity of virtual network topology and blocking probability reduced.
In this paper, the guideline of wavelength assignment for WDM distribution networks is studied and a novel wavelength assignment algorithm is proposed. The numerical and simulations results demonstrate that this algorithm can improve the system's performance.
A novel hybrid WDM/TDM optical access network architecture called wavelength shared WDM PON and its media access control mode is proposed. The system is based on a two-stage distribution tree connecting OLT and ONUs.and uses ALOHA protocol for control channel and FIFO scheduler to efficiently assign bandwidth. The packet delay in the system is analyzed and the numerical results demonstrate that the system possesses well performance of packet delay.
Recently, TCP performance over Optical Burst-Switched Networks (OBSN) has drawn more and more attention, but most of previous work were based on small linear network under static traffic, which was much different from the real condition. In this paper, a more realistic OBSN model was set up and dynamic TCP performance over it was investigated with the help of a ns2-based OBSN simulation platform. The impacts of OBSN configuration parameters and different TCP implementations on the overall throughput were discussed.
In this paper, a traffic-grooming problem for multi-granularity traffic of SDH/SONET in WDM grooming mesh networks is investigated. Our objective is to improve the throughput of SDH/SONET WDM mesh networks. We propose a heuristics algorithm to solve this problem. The performances of this traffic grooming heuristics algorithm are evaluated in WDM grooming networks. Finally, we presented and compared the simulation results of this methodology in dynamic traffic grooming WDM mesh networks with that of other methodologies.
Utilizing optical technologies for the design of packet switches and routers offers several advantages in terms of scal-ability, high bandwidth, power consumption, and cost. However, the configuration delays of optical crossbars are much longer than that of electronic counterpart, which makes the conventional slot-by-slot scheduling methods no longer be-ing the feasible solutions. Therefore, there must find some tradeoff between the empty time slots and configuration overhead. This paper classifies such scheduling problems into preemptive and non-preemptive scenarios, each has its own advantages and disadvantages, and decides whether the traffic in one input queue should be completely transmitted in one switch configuration. Although non-preemptive scheduling is inherently not good at achieving above-mentioned tradeoff, it is shown, however, that the proposed maximum weight matching (MWM) based greedy algorithm is guaranteed to achieve an approximation 2 for arbitrary configuration delay, and with a relatively low time complexity O(N2). For preemptive scheduling, a novel 2-approximation heuristic is presented. Each time in finding a switch configuration, the 2-approximation heuristic guarantees the covering cost of the remaining traffic matrix to have 2-approximation. Simulation results demonstrate that 2-approximation heuristic (1) performs close to the optimal scheduling; (2) outper-forms ADJUST[1] and DOUBLE[3] in terms of traffic transmission delay and time complexity.
In this paper, we propose a novel modified architecture of Optical Burst-Switched Networks, called Reliable Optical Burst Switching (ROBS), which introduces congestion control and retransmission mechanisms of TCP to OBS layer. Design scheme of edge and core nodes for ROBS are both demonstrated. Part of the ROBS function has been implemented on ns-2 platform. Performance improvements due to burst retransmission are simulated and presented.
In IEEE 802.17 draft standard, bridging technique is suggested to connect multiple resilient packet rings (RPRs). However, we demonstrate that bridging technology may not be a feasible solution because too much of resources and overhead would be occupied or needed to transmit packets across multiple RPR sub-networks. To construct interconnected RPRs and thus solve bridging related issues, this paper introduces a new kind of RPR nodes, namely, RPR-X nodes, to interconnect multiple RPR sub-networks. In our solutions, standard RPR frame defined by IEEE 802.17 is used to transmit packets on the local RPR sub-network, the extended RPR frame (i.e., cross frame) is used to transmit the packets across different RPR sub-networks. For RPR-X node, it can identify and process both standard RPR frame and extended RPR frame. While for RPR-S node (standard RPR node), it can only identify and process standard RPR frame. Based on the extended definitions of standard RPR frame, we give a detailed design of MAC layer reference model of RPR-X node, and propose an original packet cross forwarding algorithm, with a view to transmitting the packets across multiple RPR sub-networks, and at the same time to guaranteeing the cross forwarding algorithm to be compatible with the standard forwarding procedure of RPR-S node. Theoretical analysis and simulation results demonstrate that the proposed methods perform well in terms of performance metrics such as network's throughput and end-to-end delay.
Based on the scenario that the number of wavelengths may be fixed by upgrading the WDM metro ring networks, this paper proposes a novel WDM slotted-ring architecture with nodes employing one fixed transmitter and a set of fixed receivers. It presents a detailed design and analysis of node architecture and MAC protocol, which aim at guaranteeing fairness to each node and achieving efficient bandwidth utilization. The main advantages of this proposed node architecture and its corresponding MAC protocol can be summarized as: 1) Electronic buffer and process delays can be minimized as the packets will not be processed until they arrive at the destination nodes. 2) No contention will occur at the destination nodes. 3) No tunable components used. 4) No dedicated control channel used to transfer the packets. 5) Packets can be stripped by both destination and source nodes, and there is no need to perform O-E-O conversions. It is shown through theoretical analysis and simulation results how the proposed architecture and MAC protocol can achieve a good throughput, low queuing delay and reasonable packet loss probability. In the case of a large number of wavelengths, or if the number of wavelengths increases by upgrading the networks, this paper further proposes an original design method: employing one fixed transmitter and a set of tunable receivers that can only be tuned among partial continuous wavelengths, we show this method can achieve a good trade-off in terms of performance metrics such as network’s throughput, cost and implementing complexity.
The rapid increase of IP service and large capacity in DWDM layer request the all-optical packet delineation and process. Optical burst switching is a promising technology in the future IP/DWDM networks, which provide a feasibility to realize the optical packet switching without optical buffer. Presently the QoS guarantee of OBS depends on the bandwidth control protocol. JET based on RFD generally applied in OBS network is considered in this letter. According to the queuing theoretical model, the performance of one node is studied using simulations. Besides the traditional exponential arriving and service traffic, we analysis the general and self-similar traffic under the same average service time. The analysis and simulating results for different class have important effects on networking performance by comparing the queuing service models.
It has been proved that through burst segmenting, and thus dropping the segment of one burst that is overlapped with another burst, the packet loss probability will be significantly improved for optical burst switched networks. Based upon this, two main segment dropping strategies, head- and tail-dropping, can be used to resolve contention. In this paper, we point out that tail-dropping policy, which is adopted in most literatures, may not be a feasible solution, while head-dropping policy, even though it may make the packets to arrive out of order, is in deed a solution which can resolve the contention effectively. Moreover, with respect to tail-dropping policy, the increase of the disorder of packet arrivals for head-dropping policy is trivial. To support service differentiation and decrease the disorder of packet arrivals generated by contention resolution, we further propose a new threshold-based hybrid-assembly scheme. The most striking characteristic of the hybrid-assembly scheme is that the low and high packet classes are aggregated into one burst simultaneously. Once contention occurs, head-dropping policy is adopted to drop the overlapping packets. We describe the concrete implementation of QoS supporting, and the corresponding dropping strategy -- improved head-dropping policy, which aims at guaranteeing a better QoS support and a feasible implementation, is also detailed. Simulation results demonstrate that the proposed burst assembly scheme, together with the head-dropping policy, perform well in terms of performance metrics such as the average packet loss probability and service differentiation.
This paper focuses on the problem of optical buffer control strategies for recirculation shared optical buffer in optical packet switching. Five control algorithms are presented, and their performances are evaluated and compared by computer simulations. Issues on the optimal configuration of optical buffers are also discussed. The numerical results show that the time delay resolution of fiber delay line has significant influence on the system performance in all three discussed switching mode. The results are valuable in designing the optical packet switch.
This paper investigates the implementing techniques concerning 2-to-1 node receiver for all-optical packet switching. We point out that the state-of-the-art technologies introduced in the literatures are all associated with one main limitation, i.e., even if two incoming packets are directed to the different ports and thus no contention will occur, they can not be directed to the output ports simultaneously. To overcome this problem, a novel node receiver model is proposed in this paper. This work presents a solution that makes use of fiber delay-lines (FDLs) which allow multiple packets to be concurrently stored or transmitted. With a novel switch control, it is shown that this solution is very efficient to resolve contention, and overcomes the limitation existing in the current models (e.g., Quadro, M-Quadro and COD architectures). We describe and analyze the concrete scheduling of the switches. Simulation results based on bursty and non-bursty traffic scenarios demonstrate that the proposed node architecture (1) performs well in terms of performance metrics such as packet loss probability; (2) has a simple control requirement. In addition, we find that the required number of the recirculating time of the packets is minimal.
P-cycle based protection has been proved to be an effective way for WDM networks’ survivability design. In IP over WDM networks, p-cycle can also be constructed through GMPLS technology to protect a number of IP links. In this paper, we demonstrate that there are some drawbacks associated with the current p-cycle searching algorithms, e.g., 1) none of the them take the hop-count constraint into account, which means some requests may not able to be protected by their associated p-cycles; 2) to protect those failures occurred on the straddling links, the algorithms demand the traffic to be split in half over each p-cycle segment, which may not be a resource efficient solution, and the routing procedure may be prohibited by hop-count constraints, also, it may introduce extra link propagation delays and causes IP packets to arrive out of order. We propose an improved object function, and give a novel p-cycle searching algorithm. Simulation results obtained from CERNET-like physical topology and the national backbone networks show that the developed algorithm performs well in terms of performance metrics such as the connection blocking probability and the success rate of the protection.
In this paper, inefficiencies of the QoS scheme based on extra-offset-time-based priority for variable-length optical burst-switched network were pointed out, and a novel fixed-length burst assembly scheme is proposed. This scheme is not only practical, but also able to decrease the latency of bursts at the edge nodes. Compared to variable-length (traditional) scheme, a smaller blocking probability for the highest priority service and almost one third offset time delay for every class were obtained in fixed-length scheme.
A novel, simple and controllable bandwidth allocation scheme, called bandwidth on Demand (BoD), is proposed for optical burst-switched network in this paper. It assures every service class to consume bandwidth no more than its granted percentage by examining the bandwidth usage of each class periodically. The burst loss ratios under arbitrary input loads are theoretically analyzed as well as numerically simulated. The theoretical descriptions accord with the simulated results very well. The scheme can be used as an effective scheme of bandwidth allocation and management.
In this paper, we propose a new threshold-based mixed-assembly technique with QoS support in optical burst switched networks. The most striking characteristic of the mixed-assembly policy is that both the low and high packet classes are aggregated into one burst simultaneously. Once contention occurs, there will be an overlap between the tail of the earlier arriving burst and the head of the contending burst. Combing with burst segmentation technique, we drop the tail of the earlier arriving burst that is mainly made up of low priority packets, thus the packet loss probability of the high packet classes will be guaranteed. Simulation results show that the proposed burst assembly scheme performs well in terms of performance metrics such as the average packet loss probability and the packet loss probability of the high class of traffic.
This article focuses on the problem of switch control strategy for optical burst switch with optical buffer supporting variable length packets and under asynchronous mode. Two novel control algorithms are proposed, and their performances as well as that of the well-known reserve-switching mode, are evaluated and compared by computer simulations. Issues on the optimal configuration of optical buffers are also discussed. The numerical results show that the time delay resolution of fiber delay line has significant influence on the system performance in all three switching modes.
Holographic Transformation (HT) of a black-white bitmap image was successfully realized using the free transferring principles of light and the principle of computer generated holography. The original image was successfully reconstructed on the computer. The possibility of using HT for image encryption in image communication is experimentally analyzed and testified. It is shown that five parameters in transformation and their sequence can be taken as the key to regenerate the original image, thus a certain degree security is realized.
ATM based passive optical networks can provide broadband services as defined by the international telecommunication union. The medium access control protocol is of great importance to the access scheme as it controls the flow of traffic in the access network. This paper presents a novel MAC protocol with the ability of traffic profile distortions control for APON system at low cost of bandwidth consumption. We evaluate the protocol performance in the term of cell transfer delay and cell delay variation at different traffic load by computer simulations, which approve the validity of the protocol and the bandwidth allocation algorithm.
In high power cw-CO2-laser welding, assisting gas is necessary. It has a great effect on the welding results, such as penetration depth, cross section, size of the heat affected zone and welds surface. Much has been done on discussing the influence of its kind and flow rate on the welding results, but little attention was given to its configurations. In this article, several configurations of assisting gas were tried and the according welding results were given. Especially, a new gas assisting configuration including three gases was suggested. Their influences on the topography and cross sections were qualitatively analyzed. Experimental results showed that the best welds were obtained with the new three-gas-configuration.
The thermal deformation in penetration welding with a laser has been investigated as a factor which results in the elevation on the surface of the weld. Qualitative analysis of the course of thermal deformation during which elevation is made has been presented in detail. A series of comparison laser welding experiments was done with a 5kW CO2 laser. The experimental results showed that there were always a slight warp and a lateral shrinking of the workpiece together with the elevation on the surface, which had been predicted by the analysis. The influence exerted by shielding gas on the topography of the weld has also been discussed. The shielding gas is supposed to affect the pressure above the surface of the welding pool. The shape of the surface of the welding pool is determined by the difference between pressures on both sides of the surface in term of Laplace equation. When the distribution of the gas pressure was homogeneous within the range of the welding pool, a fairly smooth topography was experimentally obtained. Otherwise, obvious ripples were observed along the weld.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.