In this paper, a dynamic three-dimensional measurement method based on convolution neural network and binocular structured light system is proposed. We propose a convolution neural network to extract the real and imaginary terms of the first-order spectrum of a single frame fringe pattern. In our learning model, the loss function is established with output consistency, phase consistency and feature consistency as the joint constraints. And the dataset is built with actual deformed patterns of different scenes and frequencies. Furthermore, a dual frequency stereo phase unwrapping algorithm based on virtual plane is designed. Combined with the network, the absolute phase can be obtained by only two fringe projections in the measurement range, enabling the dynamic three-dimensional reconstruction of discontinuous or multiple isolated objects. The experimental results show the proposed network can significantly improve the accuracy of phase retrieve by 20 times compared to Fourier Transform Profilometry and the measurement error of the measurement system proposed in this paper for calibration sphere is less than 0.04mm. Furthermore, the measurement results of the dynamic process of palm unfolding verify the feasibility and the effectiveness of the proposed method.
The spectrogram can clearly show the composition of different frequencies in the speech signal. In this paper, a speech enhancement method based on deep learning image processing is proposed, which optimizes the spectrogram of the laser detected speech signal to achieve speech enhancement. The laser beam emitted by the laser Doppler vibrometer (LDV) is focused on the glass window to detect the vibration caused by sound wave. After conversion, the audio information that causes vibration is obtained. Under the interference of speckle noise and air disturbance, the detected speech signal not only has a low signal-to-noise ratio (SNR) but also has non-stationary noise. In order to overcome the difficulty that traditional methods are difficult to extract weak signals in the case of severe noise interference, we use deep learning to achieve spectrogram noise reduction and speech information enhancement. By processing the spectrogram of noisy speech with the generative adversarial networks (GAN) combined with the spatial attention mechanism and introducing the short-time objective intelligibility (STOI) into the loss function, the laser detected speech signal was successfully enhanced.
Optical encryption technology using phase-shifting interferometry (PSI) is widely used in many security fields, showing the advantages of parallel processing and the high security due to the information can be encrypted in multiple dimensions. The traditional PSI- based encryption methods are realized by the traditional multi-step phase-shifting algorithms (PSAs), although they are fast and accurate, the phase shifts deviation (PSD) caused by phase shifter detuning often affect the performance of the above methods. In order to solve this problem, a phase-only encryption method combining a normalization and orthogonalization phase-shifting algorithm (NOPSA) is proposed to eliminate the influence of PSD on the effective on decryption. First, the original image is phase encoded and encrypted by a phase key loaded on a liquid crystal spatial light modulator (SLM). Subsequently, the modulated reference light will be overlapped with the object wave to form interference pattern. Finally, the encrypted image hiding can be implemented based on NOPSA. Compared with the current PSI- based encryption methods, its obvious advantage is that when the phase shifts are not accurate or unknown, the proposed method can still achieve high quality encryption and hiding, which broadens the working conditions of PSI- based encryption methods.
Calibration which defines the relationship between the phase and depth data is the important part of the fringe projection
profilometry. In practice, the inherently nonlinear and spatially variable relationship between the absolute phase of the
projected fringe and the object surface depth without using telecentric lens make calibration problematic in the
measurement of small object. In order to obtain this problem, a flexible, simple telecentric three dimensional
measurement system is proposed. Because of the characteristic that the size of object will not change with depth in
telecentric imaging, the absolute phase is linear with the depth and the process of calibration become simpler. The
experiment result indicate that the standard deviation of calibration result at z coordinate is within 5 μm, while that at x
and y coordinate is within 3 μm. Three-dimensional shape reconstruction of a coin value ¥1 and measurement of central
circle points of the calibration target further verify the validity of the proposed calibration method.
KEYWORDS: Structured light, Digital signal processing, 3D metrology, Embedded systems, Binary data, Modulation, Cameras, Imaging systems, Projection systems, 3D modeling
Structured light measurement has been wildly used since 1970s in industrial component detection, reverse engineering, 3D molding, robot navigation, medical and many other fields. In order to satisfy the demand for high speed, high precision and high resolution 3-D measurement for embedded system, a new patterns combining binary and gray coding principle in space are designed and projected onto the object surface orderly. Each pixel corresponds to the designed sequence of gray values in time – domain, which is treated as a feature vector. The unique gray vector is then dimensionally reduced to a scalar which could be used as characteristic information for binocular matching. In this method, the number of projected structured light patterns is reduced, and the time-consuming phase unwrapping in traditional phase shift methods is avoided. This algorithm is eventually implemented on DM3730 embedded system for 3-D measuring, which consists of an ARM and a DSP core and has a strong capability of digital signal processing. Experimental results demonstrated the feasibility of the proposed method.
A method is presenting for measuring beam propagation ratios (M2) of laser beam utilizing transmissive liquid crystal spatial light modulator (LC-SLM). In this paper, the function of digital zoom lens (DZL) could be obtained by loading a calculated gray-scale of DZL with desired focal length into the LC-SLM and the accurate focal length of DZL could be gotten by a correction method. By comparing with the standard measuring method provided by ISO (International Standard Organization), through using the transmissive LC-SLM, the beam propagation ratios could be calculated by curve fitting based on the beam spot size collected by digital camera versus the focal length of DZL without any movements in the experiments. The experiment and comparison results have shown that it is effective for us to measure the M2 of Gaussian beam using a transmissive LC-SLM whose experimental setups are simple, relatively.
Binocular Vision Technique is widely used in three-dimensional(3-D)measurement. Matching of pictures captured from two cameras is the most critical and difficult step in 3-D shape reconstruction. The method combines codedstructured light and spatial phase is usually adopted. However, being time consuming in matching, this method could not meet the requirements of real-time 3-D vision. In order to satisfy the high speed characteristic of real-time measurement, a novel method using gray level vector modulation is introduced. Combining binary code with gray coding principle, new coding patterns using gray level vector method is designed and projected onto the object surface. Each pixel corresponds to the designed sequence of gray values as a feature vector. The unique gray level vector is then dimensionally reduced to a resulting value which could be used as characteristic information for binocular matching. Experimental results further demonstrated the correctness and feasibility of the proposed method with fewer component patterns and less computational time.
KEYWORDS: Image registration, Range image registration, 3D image processing, Cameras, 3D modeling, Data modeling, Algorithm development, 3D metrology, 3D image reconstruction, Image processing
This paper addresses the range image registration problem for views having overlapping area and which may include substantial noise. The current state of the art in range image registration is best represented by the well-known iterative closest point (ICP) algorithm and numerous variations on it. Although this method is effective in many domains, it nevertheless suffers from two key limitations: It requires prealignment of the range surfaces to a reasonable starting point and it is not robust to outliers arising either from noise or low surface overlap. This paper proposes a new approach that avoids these problems for precision range image registration, by using a new, robust method based on ASIFT followed by ICP. Up to now, this approach has been evaluated by experiment. We define the fitness function to calculate the time for the convergence stage of ICP, because the time required is very important. ASIFT are capable of image matching even when there is fully affine variant. The novel ICP search algorithm we present following ASIFT offers much faster convergence than prior ICP methods, and ensures more precise alignments, even in the presence of significant noise, than mean squared error or other well-known robust cost functions.
The traditional structured light binocular vision measurement system consists of two cameras and a projector, which can
be regarded to two monocular vision systems composed by the projector and a camera. In this paper, we present a threedimensional
(3D) measurement method based on the combination of binocular vision and monocular vision. The
common field of view is reconstructed by a binocular vision system, and the missing data area is filled up by two
monocular vision systems. In order to improve the measurement accuracy and unify the three world coordinate systems,
a calibration method is proposed. The calibration procedure consists of a binocular vision system calibration, the two
monocular vision systems calibration and a globe optimization of the three systems for unifying to a common reference.
In monocular vision system calibration, a new method based on virtual target is proposed and used to set up the
coordinate relations. We use a projector and two cameras to build a vision system for testing the proposed technique. The
experimental results show the calibration algorithm ensures the consistent accuracy in the three systems, which is
important for data fusion. And it is clear that the proposed method improves the integrity of measurement results and
measuring range efficiently.
Nondiffracting beams are of interest for optical potential applications owing to their properties of smaller central spot,
longer propagation distance and so on. A phase-holograms-based method of generation of nondiffracting beams array
with arbitrary order is proposed in this paper. If a phase hologram is displayed on a phase-only spatial light modulator
(SLM), when a collimated monochromatic plane wave illuminates it, an annulus with high concentration of energy is
obtained in the Fourier plane. Then through the Fourier transform again, a nondiffracting beam will be generated. It is
able to generate arbitrary order nondiffracting beams with high diffractive efficiency. More significantly, if a phasehologram-
array that possesses the same eigenvalue is utilized, a unique bright annulus will be generated in the Fourier
plane because of the shift-invariance of Fourier transform and the consistency of phase hologram design. Then through
the Fourier transform again, a nondiffracting beams array will be generated. Furthermore, the location and the order of
each individual nondiffracting beam can be customized according requirement. Experiment results are in good
agreement with the numerical simulation and the theoretical analysis.
A method for improving the performance of phase-only spatial light modulator (SLM) is proposed in this paper. For an
electrical-addressed SLM, the pixelated structure just likes a 2D black-matrix. It will have an intrinsic effect on the
incident light whether image is loaded into SLM. This inherent effect was called black-matrix effect characterized with
zero-order diffraction of high brightness and high-order diffraction terms in the Fourier plane. This is strongly influenced
the quality of phase modulation. In order to eliminate the black-matrix effect of SLM, a linear phase map or a digital
Fresnel lens is compensated to the original phase map. The black-matrix effect will be separated with the reconstruction
pattern in the perpendicular plane of the optical axis or along the optical axis. Therefore, the black-matrix effect will be
eliminated through digital phase compensation. The theoretical analyze, computer simulations and experimental results
are all presented to demonstrate the validity. Possible applications include almost all phase modulation applications.
In this paper, a method of measuring the phase modulation properties of spatial light modulator (SLM) by heterodyne
interferometry is proposed. As a kind of key elements in the advanced optical information processing systems, spatial
light modulators is widely used in many important fields, especially used as a dynamic phase modulating device. So, the
phase calibration plays an important role in the SLM applications. Compared with the methods based on traditional
interferometry, this method measures the phase response directly by taking advantage of the heterodyne mechanism, so it
leads to higher accuracy. A heterodyne-interferometer based calibration system making use of acousto-optic frequencyshifters
has been designed and realized. Theoretical analysis and experimental results demonstrate the validity of this
method.
KEYWORDS: 3D metrology, Cameras, Fringe analysis, Projection systems, 3D modeling, 3D image processing, Imaging systems, Computer programming, Phase shifts, RGB color model
A high-resolution, dynamic Three-dimensional (3-D) profilometry based on the combined stereovision and color-encoded digital
fringe projection is proposed. In this technique, a sinusoidal fringe pattern is encoded with spatial neighborhood strategy based on De
Bruijn sequences. A decoding algorithm for the color pattern is presented. The absolute phase value is retrieved by space method
based on locally intensity variety, and unwrapped by dividing the periods based on the intensity peak and the corresponding color
information. Therefore, only a single color image is needed to realize the unique code in pixel dimension, which meets the demand of
high-resolution, real-time 3D shape measurement. That means this technique could realize pixel-level resolution and measure
disconnected objects. Since the phase value at each pixel is only used to assist stereo matching, the 3-D reconstruction could be realtime,
and the accuracy is also enhanced. A measurement system consisted of one projector and two cameras is developed.
Experimental results are presented to show the feasibility of the proposed method.
KEYWORDS: Fringe analysis, Projection systems, 3D metrology, Digital signal processing, Phase shifting, Phase shifts, Embedded systems, Cameras, Digital Light Processing, Imaging systems
An embedded three-dimensional (3-D) profilometry system based on a combination of gray-code and phase shifting (GCPS) method
is proposed. This system consists of a digital-micromirror-device (DMD) based video projector, a high-speed CCD camera and an
embedded digital signal processing hardware system based on DSP. In this technique, seven gray-code patterns and three sinusoidal
fringe patterns with 120-deg phase shift are integrated in red, green and blue channels to form four color fringe patterns. When the
four color fringe patterns are sent to the DMD based projector without color filter, the previous gray-code patterns and three sinusoidal
fringe patterns are repeatedly projected to an object surface in gray-scale sequentially. These fringe patterns deformed by the object
surface are captured by a high-speed CCD camera synchronized with the projector. An embedded hardware system is developed for
synchronization between the camera and the projector and taking full advantage of DSP parallel processing capability for real-time
phase retrieve and 3-D reconstruction. Since the number of projected images of GCPS is reduced from 11 to 4, the measurement speed
is enhanced dramatically. Experimental results demonstrated the feasibility of the proposed technique for high-speed 3-D shape
measurement.
As a crucial part of active three dimensional(3D) vision system based on fringe projection technique, correspondence
search between two adjacent range images would directly influence on the accuracy of matching and fusion procedure of
the depth data. The conventional sub-pixel matching method by means of phase correlation can achieve a high level of
accuracy while it would also be associated with a time consuming procedure due to the requirement of capturing two
orthogonal series of sinusoidal fringes. Another technique utilizing linear interpolation algorithm based on the fringe
projection with single direction might cause accuracy declination as a result of the simplification in imaging model. A
novel method under the framework of linear interpolation was proposed. This approach makes use of absolute phase
values and parameters of epipolar line as two kinds of feature points for correspondence search, leading to a significant
improvement on the measurement accuracy of the depth data. Theoretical analysis and experiment results demonstrate
the validity of presented approach.
A generalized temporal phase unwrapping (GTPU) algorithm is proposed for absolute phase measurement of object
surfaces with complex topography. Comparing with the classical TPU algorithms, GTPU shows more robust and has
better noise immunity and less computational complexity. According to the type of fringe sequence used, the current
TPU algorithms proposed by Huntley can be divided into three categories: linear sequence, exponential sequence and
reversed exponential sequence. The two types of exponential sequences make use of the fact that the relation between the
phase and the number of projected fringes is linear, resulting in a reduction of total number of acquired images compared
to linear sequence illumination. Error analysis for the different TPU methods were done firstly, that reveals the problems
existed in theses phase unwrapping methods, for example the limitation imposed on the fringe sequence, noise immunity
and computational efficiency. In order to overcome these drawbacks, we present a generalized TPU algorithm in this
paper. The GTPU can eliminate the limitation on fringe sequence, so that an arbitrary fringe sequence will be used to
encode the object surface, leading to a flexible method for phase reconstruction. Computer simulations and experiment
results have been also proved that the GTPU has better performance on noise control and computational efficiency.
A novel method for the coarse registration of range images is proposed. This approach is based on texture-feature
recognition. As the development of optical digitizing technique, it is now able to acquire the range images and associated
texture images sequentially or simultaneously. It's possible to identify the range feature points through texture feature
points. Scale Invariant Feature Transform (SIFT) is an efficient method for texture feature generation. SIFT transforms
texture image into a large collection of local feature vectors, each of which is invariant to image scaling, translation, and
rotation. The mismatched correspondence pairs can be discarded using random sample consensus algorithm based on
epipolar geometry constraint. We select more than three well-registered texture-feature pairs, with which we could find
the associated range-feature pairs of the range images. Initial pose estimation of the two involved range images can be
computed by these range pairs, and the fine registration is implemented using iterative closest point (ICP) algorithm. Our
approach utilizes the texture information to register the range images, leading to a technique that can be automatically
performed while the influence of 3D noise can be avoided. The experiment results demonstrate that the proposed
approach is efficient and robust for the registration of multiple range images.
KEYWORDS: Digital signal processing, Fringe analysis, 3D image processing, Signal processing, Phase shifts, Digital imaging, Imaging systems, Video, 3D modeling, 3D metrology
An embedded three-dimensional (3-D) digital imaging scheme with a parallel fixed-point digital signal processor (DSP) is presented, which is based on temporal phase unwrapping and DLP digital projection technique. This scheme utilizes an embedded hardware structure with aid of parallel DSP to realize a pipeline procedure of the automatic analysis of fringe patterns, fringe patterns generation, fringe projection, and data acquisition. The software pipeline is also adopted in the procedure phase demodulation and phase unwrapping. The time of phase reconstruction with five-step phase shift is 0.89s for 262144 coordinates, and the fringe image processing speed is up to 39.5 f/s (frames per second), so it can meet the need of video-rate fringe images processing. Experiment results show that this embedded DSP imaging system is fast, reliable, low power cost, and it will be suitable for a wide range of practical measurement application.
With the improvements in range image acquisition by optical metrology of our group, we also developed a novel method
for the registration and integration of range images. The registration approach is based on texture-feature recognition.
Texture-feature pairs in two texture images are identified by cross-correlation, and the validity-checking is implemented
through Hausdorff distance comparison. The correspondence between the texture image and range image helped acquire
the range point-pairs, and the initial transformation of two range images was computed by least-squares technique. With
this initial transformation, the fine registration was achieved by ICP algorithm. The integration of the registered range
images is based on ray casting. An axis-aligned bounding box for all range images is computed. Three bundles of
uniform-distributed rays are cast and pass through the faces of the box along three orthogonal coordinate axes
respectively. The intersections between the rays and the range images are computed and stored in Dexels. The KD-tree
structure is used to accelerate computation. Those data points in overlapped region are identified with specific criteria
based on the distance and the angle of normals. We can obtain a complete non-redundant digital model after removing
the overlapped points. The experimental results illustrate the efficiency of the method in reconstructing the whole three dimensional
objects.
We report an approach for three-dimensional (3D) imaging with use of acousto-optic fringe projector (AOFP) and a
technique of piecewise temporal phase unwrapping. The AOFP is controlled by direct digital synthesizer to generate a
sequence of fringe patterns with different spatial frequencies so that encoding with variable sensitivity can be realized.
Furthermore, we present a new phase unwrapping strategy for incremental phase reconstruction with the level of details.
The algorithm is developed as a piecewise temporal phase unwrapping with a tolerance of deviation from the condition
imposed upon conventional temporal phase unwrapping technique. Preliminary experiment results are given to support
proposed approach.
X-ray phase-contrast imaging is an important diagnostic tool in medicine, biology and materials science. In-line hard x-ray phase-contrast imaging is based on Fresnel diffraction of x-ray, therefore we propose to make phase retrieval calculations between arbitrary planes interrelated through the Fresnel domain. A new approach to the numerical reconstruction of object phase by the diffraction intensity for in-line x-ray phase-contrast imaging is presented. The new method is tested on simulated image and the results demonstrate the validity of this new approach.
A cascaded iterative angular spectrum approach (CIASA) based on the methodology of virtual optics is presented for optical security applications. The technique encodes the target image into two different phase only masks (POM) using a concept of free-space angular spectrum propagation. The two phase-masks are designed and located in any two arbitrary planes interrelated through the free space propagation domain in order to implement the optical encryption or authenticity verification. And both phase masks can serve as enciphered texts. Compared with previous methods, the proposed algorithm employs an improved searching strategy: modifying the phase-distributions of both masks synchronously as well as enlarging the searching space. And with such a scheme, we make use of a high performance floating-point Digital Signal Processor (DSP) to accomplish a design of multiple-locks and multiple-keys optical image encryption system. An evaluation of the system performance is made and it is shown that the algorithm results in much faster convergence and better image quality for the recovered image. And two masks and system parameters can be used to design keys for image encryption, therefore the decrypted image can be obtained only when all these keys are under authorization. This key-assignment strategy may reduce the risk of being intruded and show a high security level. These characters may introduce a high level security that makes the encrypted image more difficult to be decrypted by an unauthorized person.
In order to implement 3-D imaging of objects with large height discontinuities and/or surface isolation, we present a novel 3-D imaging system based on temporal sequential fringe projector to provide multi-resolution 3-D reconstruction. To recover the range data of such a surface, an enhanced scheme for temporal phase unwrapping procedure is proposed. We also describe methods for extracting the color texture corresponding to a range image. Experimental results are given to illustrate the validity of our proposed method.
A three-dimensional (3D) profilometry based on a dual-acousto-optic fringe projection is reported in this paper. The fringe projector is able to generate fringe patterns with different optical sensitivities and therefore can handle the object surface with complex geometry and topology. It is also qualified for video-rate 3-D profilometry of arbitrary shape objects. By using two acousto-optic deflectors (AODs), the projector can generate a time-series, frequency-varying fringe patterns by use of the AO interaction effect. The two AODs are driven by the radio frequency (RF) signal with the same frequency. Changing the RF frequency leads to the change of the period of fringe patterns and, therefore, different sensitivities. In practice, the changing rate of sesitivity is limited by the frame rate of the detector, e.g. CCD, so it makes a dynamic 3-D profilometry possible. Experimentally, we project a sequence of spatial frequency-varying fringe patterns onto the test object surface. Then, a CCD camera simultaneously acquires the phase-modulated fringe patterns accordingly. With a sequence of phase measurement in a progression of fringe periods, an enhanced scheme for phase unwrapping algorithm is implemented in a recursive manner. The object surface with complex geometry and topology can be reconstructed in this way. Experimental results are also given to validate this approach.
We report on the development of a multispectral multiphoton fluorescence lifetime imaging microscopy (MM-FLIM) system that is the combination a streak camera, a prism spectrophotometer, a femtosecond Ti: Sapphire laser and a fluorescence microscope. This system is versatile with multispectral capability, high temporal (10ps) and spatial (0.36μm) resolution and can be used to make 3-dimensional (3D) (x-y-z) multiphoton fluorescence intensity, spectrally resolved intensity and lifetime measurements with a single detector. The system was calibrated with a F-P etalon and a standard fluorescent dye and the lifetime value obtained was in good agreement with the value reported in the literature. Preliminary results suggest that this MM-FLIM system has integrated high temporal, spatial, and spectral resolution fluorescence detection in one microscopy system. Potential applications of this system include multiwell imaging, tissue discrimination, intracellular physiology and fluorescence resonance energy transfer imaging.
This paper presented an optical interconnection multiple PC system connected by a ring network. This multiple PC system uses Linux Operating system, and the standard communication protocol of complying with Ethernet specification. This paper studied the factors that influence the interconnection performance among multiple computers, including the effect of the buffer size of OLI on the bandwidth of network layer, and the effect of retransmission method on the performance. The results of ping-pong test show that optical fiber link is able to provide high-bandwidth and low-latency communication for a multi-PCs system. We found that the performance of the interconnection network can be improved by improving the interface match between the optical fiber link and computer bus input/output. Besides, the development of a chip to implement network interface and a part of network protocol, such as developing with FPGA device in this paper, could result in the higher performance of the interconnection.
KEYWORDS: Data transmission, Field programmable gate arrays, Clocks, Human-machine interfaces, Data communications, Interfaces, Local area networks, Switching, Computer networks, Time division multiplexing
A giga-bit per second optical ring has been designed. The data link layer bandwidth of the network is 1.5 Gbit/s. Hardware routing has been achieved by field programmable gate array (FPGA) to minimize the communication latency. As a result, the point-to-point communication latency between adjacent node computers is less than 300 ns. To reduce the data communication latency and have a more efficient usage of the data link layer bandwidth, virtual multi-channel transmission mechanism has been achieved with hardware. The network supports a maximum of four virtual channels for each physical data link. Time slots can be dynamically assigned to each virtual channel. The length of each time slot can also be dynamically selected based on the length of each data packets. Zero wait time has been achieved for the switching between different virtual channels. The relationship between the network performance and the number of virtual channels has also been analyzed in this paper.
In this paper, a communication protocol has been designed for the two-layer scalable optical interconnection network (TSOINet) designed in Tianjin University. The communication protocol has been designed based on TCP/IP (transmission control protocol/internet protocol) protocol. But it is more efficient than TCP/IP for data transmission in TSOINet. An auxiliary channel has been introduced in TSOINet for transmission of the communication status messages among the sub-layer rings. Many network management functions, such as flow control, hand shaking, etc., can be achieved via the auxiliary channel, so the communication overheads can be decreased. The performance of the communication protocol has been evaluated compared to TCP/IP protocol for data transmission in TSOINet.
In this paper, some study results to apply fiber link to a computer cluster are presented. The research is based on a ring network topology for a cluster system, which is connected by gigabit/s virtual parallel optical fiber link (VPOFLink) and its driver is for Linux Operating System, the transmission protocol of VPOFLink is compliant with Ethernet standard. We have studied the effect of different types of motherboard on transmission rate of the VPOFLink, and have analyzed the influence of optical interconnection network topology and computer networks protocol on the performance of this optical interconnection computer cluster. The round-trip transmission bandwidth of the VPOFLink have been tested, and the factors that limit transmission bandwidth, such as modes of forwarding data packets in the optical interconnection ring networks, and the size of the link buffer etc., are investigated.
In this paper, we present a low cost fiber-optic data link for the computing cluster. The link is a 32 bit-virtual- channel fiber-optic computer bus used only a pair of OE devices and fibers. The link is integrated with the popular PCI Bus interface in order to make the link hold the same bandwidth as that of the PCI Bus, and it can operate under managed by PCI Bus. Our research addresses how to match accessing bandwidth between computer bus and high-speed optical interconnections, and how to make low overhead and latency interfaces between optics and electronics, and integrates fiber optic link into computing system designs.
This paper describes the two-layer scalable wavelength routing optical interconnection network being developed in Tianjin University. The top layer of the network is multi- wavelength bi-directional optical bus, which has high bandwidth and low latency. The optical bus is made up of passive components, no wavelength-tunable devices have been sued. As a result, the optical bus has low communication latency that is mainly decided by the optical fiber length. The sub-layer of the network is single-wavelength ring, which has low communication latency and high-scalability. In each ring, a wavelength routing node is used for data transmission between the ring and the optical bus. Each node computer is connected to the ring using an optical network interface card, which is based on peripheral component interconnect bus. The communication latency inside the ring is decreased using synchronous pipelining transmission technique. The scale of the ring is mainly limited by the efficient bandwidth required by each node computer. The number of rings is mainly decided by the optical power of the laser diodes and the sensitivity of the optical detectors. If Erbium doped fiber amplifier is used in the optical bus, the scale of the network can be further developed.
In this paper, wavelength routing technology is applied to computer interconnection network. By analyzing the relationship between wavelength and networks routing, we describe a concept of wavelength being used as network address, and propose a wavelength routing topology to extend the scale of networks and realize scalability of networks. A twin-wavelength ring network to implement and test the function of wavelength routing is presented, and the main units of the twin-wavelength ring network are presented as well. In addition, we put forward tow methods to implement it. The design method of physical link driver software for Linux, which uses optical interconnection interface as network interface for parallel computing, is also introduced.
It is well known that a wavelength can be regarded as an address code of a packet in the optical interconnection network. In our scheme, we use wavelength as a part of an address to build a multiple wavelengths optical interconnection ring network. The network consists of double layer rings, the routing inside a ring is electronic, and wavelength routing technology is applied to switch between rings. Two wavelength are used in the network, one is for transmitting data inside local ring, another is for switching to another ring. The detail of the network is described in this paper. Besides, a low-cost virtual parallel optical link and optical interconnection interface, which is bound up with the multiple-wavelengths network, is presented.
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.