Holographic storage device

Abstract

To provide a holographic data storage system and a method of exposing an area to a pattern over a sufficient sustaining duration (depending on an output of a beam).SOLUTION: In a method of performing a write operation in a holographic data storage system, a scheduler schedules at least one write operation across several discontinuous write intervals, and, in each of the discontinuous write intervals, a plurality of discontinuous write intervals, in which a region of a holographic recording medium is exposed to an interference pattern resulting from interference between a reference beam and an input beam carrying a set of data has a total duration sufficiently long to cause a sustained state change in the exposed region, thereby the set of data can be recovered from the region by the end of the last write interval of the plurality of discontinuous write intervals.SELECTED DRAWING: Figure 8A

Description

Technical Field

[0001] Technical Field

[0001] This disclosure generally relates to holographic memory devices.

Background Art

[0002] Background

[0002] A holographic memory device is in the form of a computer memory device, and information is recorded on a photosensitive holographic recording medium by exposing the medium to an optical pattern. For example, an area (subvolume) of the medium can be exposed to an optical interference pattern that results from the interference of an input beam and a reference beam in which a set of data is embedded. The beam can be, for example, a laser beam generated using a single laser and a beam splitter. Spatial light modulation (SLM) can be used to embed a set of data in the input beam (e.g., an image encoding the set of data can be spatially modulated and embedded in the input beam). To avoid misunderstanding, in this specification, the terms "light", "optical" and the like are not limited to visible light. A holographic memory device can be implemented, for example, using an infrared or ultraviolet beam within the non-visible portion of the electromagnetic spectrum.

[0003]

[0003] With sufficient beam output and exposure time, the optical interference pattern causes a persistent state change within the subvolume (at this point, the interference pattern is said to be persistently recorded or written to the subvolume in this specification). The state change of the subvolume, at a subsequent point in time, upon exposure of the subvolume to a substantially matching reference beam, allows the set of data originally embedded in the input beam to be restored from the output beam by the interaction between the matching reference beam and the subvolume, such that an output beam that is essentially identical to the original input beam is generated (this can be referred to as reading the recorded pattern in this specification).

[0004] Rather than storing individual bits as discrete units, a single interference pattern can encode many (e.g., millions) of bits. For example, a set of data can be a megapixel image embedded in an input beam. Moreover, by taking advantage of a certain form of sensitivity of the holographic recording medium to small changes in the angle of the reference beam, many (e.g., hundreds or thousands) of such patterns can be recorded in the same subvolume. In the case of such a medium, when an interference pattern is generated with a reference beam at a given angle, the recorded pattern can be read only by using a reference beam that is precisely aligned with the reference beam originally used to generate it. This effect can be utilized to record multiple patterns (encoding different data sets) in the same subvolume at different reference beam angles. Theoretically, the data storage capacity is limited only by the wavelength of the beam and potentially can be hundreds of megabytes per cubic millimeter for red light and dozens of gigabytes for ultraviolet light. In practice, there may be other limiting factors, but nevertheless, there is great potential for high-density data storage. SUMMARY OF THE INVENTION

[0005] Summary

[0005] This summary is provided to introduce a selected group of concepts in a simplified form; these concepts are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. Also, the claimed subject matter is not limited to implementations that solve any or all of the disadvantages described herein.

[0006]

[0006] The present disclosure relates to scheduling write operations in a holographic data storage system in an energy-efficient manner. Herein, a write operation relates to a set of data, such as an image, that is carried by an input beam and recorded in a holographic recording area as an interference pattern that results from the interference between the input beam and a reference beam. In order to store the data in a recoverable form, i.e., so that it can be read from the holographic recording area at a later time, it is necessary to expose the area to the pattern for a sufficient duration (depending on the output of the beam).

[0007]

[0007] Surprisingly, by “staggering” the write operations over a plurality of discontinuous write intervals, i.e., by exposing the holographic recording medium to the interference pattern in each of the discontinuous write intervals as described above (a “pulsed” write), except for intervening time intervals that separate the discontinuous write intervals, it has been found that the total exposure time required to record the pattern is less than the total time that would be required if the area were exposed to the pattern in a single continuous write interval (a “continuous” write).

[0008]

[0008] A first aspect of the present specification provides a method of performing a write operation in a holographic data storage system. The schedule schedules at least one write operation over a plurality of discontinuous write intervals, the write operation relating to a set of data stored in an area of a holographic recording medium. In each of the discontinuous write intervals, the area of the holographic recording medium is exposed to an interference pattern that results from the interference between a reference beam and an input beam that carries the set of data. The plurality of discontinuous write intervals have a total duration that is long enough to cause a persistent state change in the exposed area, whereby the set of data is recoverable from the area by the end of the last write interval of the plurality of discontinuous write intervals.

[0009] Brief Description of the Drawings For a better understanding of the present disclosure and to show how embodiments of the present disclosure may be implemented, reference is made to the following figures by way of example only.

Brief Description of the Drawings

[0010]

Figure 1A

[0010] Shows a schematic perspective view of a holographic recording medium.

Figure 1B

[0010] Shows a schematic perspective view of a holographic recording medium.

Figure 2A

[0011] Shows a schematic perspective view of a holographic storage system including a set of waveguides that can be used to direct beams to / from different sub-volumes of a holographic recording medium to provide spatial multiplexing across the medium.

Figure 2B

[0011] Shows a plan view of the system during the writing interval.

Figure 2C

[0011] Shows a side view of one side of the system during the writing interval.

Figure 2D

[0011] Shows a side view of one side of the system during the writing interval.

Figure 2E

[0011] Shows a plan view of the system during the reading interval.

Figure 2F

[0011] Shows a side view of one side of the system during the reading interval.

Figure 2G

[0011] Shows a side view of one side of the system during the reading interval.

Figure 3A

[0012] Shows a schematic side view of an active light pipe of a certain configuration.

Figure 3B

[0012] Shows a schematic side view of an active light pipe of a certain configuration.

Figure 3C

[0012] Shows a schematic side view of an active light pipe of a certain configuration.

Figure 3D

[0012] Shows a schematic side view of an active light pipe of a certain configuration.

Figure 3E

[0012] Shows a plan (cross-sectional) view of the active light pipe.

Figure 3F

[0012] Shows a plan (cross-sectional) view of the active light pipe.

Figure 4A

[0013] Shows a side view of one side of the optical waveguide network (a part thereof).

Figure 4B

[0013] Shows a side view of one side of the optical waveguide network (a part thereof).

Figure 5

[0014] Shows a schematic diagram of an example of a multi-waveguide network used for multiplexing across multiple pieces of a holographic storage medium.

Figure 6A

[0015] Shows an example of an emission system for providing input and reference beams in a holographic storage system.

Figure 6B

[0015] Shows a modified form of an emission system with a simplified optical system.

Figure 7

[0016] Shows an example of a data recovery system that measures the optical field of an output beam using spatial coherent detection and reduces waveguide distortion in the measured optical field using signal processing.

Figure 8A

[0017] Shows a functional block diagram showing the functions performed within a holographic storage system.

Figure 8B

[0018] Schematically shows a scheduling scheme for performing the writing operation step by step.

Figure 9

[0019] Shows an alternative holographic storage system that spatially multiplexes two-dimensionally across a slab of a holographic recording medium using at least one waveguide network.

Figure 10A

[0020] Shows how spatial multiplexing can be achieved using passive guiding elements, and spatial multiplexing is achieved by modulating beam characteristics.

Figure 10B

[0020] Shows how spatial multiplexing can be achieved using passive guiding elements, and spatial multiplexing is achieved by modulating beam characteristics.

Figure 11A

[0021] A light pipe with a passive optical filter having different frequency responses is shown.

Figure 11B

[0021] A light pipe with a passive optical filter having different frequency responses is shown.

Figure 12

[0022] An example of a waveguide network having three hierarchical levels is shown.

Figure 13

[0023] A graph of diffraction efficiency results versus writing time in pulsed and continuous writing modes is shown.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Detailed description of examples of embodiments

[0024] Embodiments implementing the above scheduling techniques are described below. First, some useful context is provided. The described embodiments implement scheduling techniques in a holographic storage system that uses a waveguide network to facilitate spatial multiplexing across one or more holographic recording media without requiring mechanical operations of the media or devices used for reading and writing therefrom. While this combination has additional advantages, it is important to note that this scheduling technique can be implemented in any holographic storage device situation.

[0012]

[0025] The described embodiments combine scheduling techniques with waveguide architectures to achieve efficient spatial multiplexing across one or more holographic storage media in a way that also reduces or eliminates the need for mechanical operations, and can use "active" light pipes, "passive" light pipes, or a combination of active and passive light pipes. Note that the terms "waveguide" and "light pipe" are used interchangeably herein.

[0013]

[0026] "Active light pipe" refers to a waveguide having one or more active switching elements or other guiding elements attached to the surface of the waveguide or within the bulk of the waveguide, and thus is capable of (i.e., having changeable optical properties) causing "one-to-many" light transmission (i.e., light is guided from a first surface region to one of a plurality of possible second surface regions, where the first surface region is the in-coupling region and the second region is the out-coupling region) or "many-to-one" light transmission (i.e., light is guided from any one of the second surface regions (here the in-coupling region) to the same first surface region (here the out-coupling region)). The term "passive light pipe" refers to a light pipe having guiding elements with different light sensitivities (e.g., different wavelength and / or polarization sensitivities), and instead, similar effects can be achieved by changing the optical properties or the beam (e.g., using an adjustable laser to change its wavelength, polarization, etc. so that it is guided along different routes by guiding elements having different wavelength / polarization responses, etc.). The term "passive light guiding" is merely a convenient label and is consistent with the fact that in this case the guiding elements do not need to be active, however, in this context, active guiding elements or passive guiding elements having different light sensitivities (e.g., different wavelengths and / or polarization sensitivities, etc.) can be used (i.e., the guiding elements can be both active and those having different light sensitivities).

[0014]

[0027] A digital image (or data encoded as a digital image) can propagate as a beam along an active or passive light pipe. The guiding elements of the active light pipe can be individually controlled to transmit or reflect the incident light beam.

[0015]

[0028] Many such light pipes (active, passive, or a combination of both types) can be combined in various geometries to create a switching network that can be used to manipulate beams and images to one of many addressable locations in one or more spatial dimensions. Inputs to the light pipe can be generated using a spatial light modulator (SLM), and outputs can be read, for example, in a CCD (charge-coupled device). Phase coherence and noise can be corrected using a combination of optical and computational techniques (including machine learning techniques). Certain embodiments use coherent detection in combination with such techniques to provide more effective waveguide distortion reduction.

[0016]

[0029] In contrast to the types of optical switches and fibers conventionally used in optical data communication, the described embodiments use a light pipe that can transmit an entire image at once. This takes advantage of currently available high-resolution optical devices such as SLMs and digital cameras. These devices have millions of pixels and enable the encoding and decoding of megabytes of data. This enables high-bandwidth transmission even at a modest switching rate of the SLM, camera, and active light pipe elements (in the case of an active light pipe) or beam optical properties (in the case of a passive light pipe). In addition, applications such as holographic storage devices require interference between multiple beams, and at least one of the multiple beams is modulated with an image. In a holographic storage device, an active light pipe can be used to efficiently manipulate the beams and images to interfere at any desired location on the holographic storage medium. As described, simpler advantages can be achieved with a passive light pipe, in which switching is instead applied at the emitter stage.

[0017]

[0030] The light pipes described below are "multimode" waveguides in the sense that they have sufficient physical dimensions to support a wide range of "modes" (i.e., spatial paths through the waveguide for a given channel, e.g., corresponding to different propagation directions). This is in contrast to simple single-mode optical waveguides such as thin optical fibers used in optical fiber systems, whose purpose is to essentially limit the light entering the fiber to a single propagation mode. Single-mode optical waveguides can only carry data using amplitude, phase, or frequency modulation, while multimode optical waveguides can carry more data (e.g., potentially an entire image of millions of pixels) by means of angular variations within the waveguide. Put another way, multimode waveguides provide a larger bandwidth through an increase in angular and / or spatial diversity by providing multiple optical paths (different paths corresponding to different propagation modes) through the waveguide from the emitter to the detector for any given channel.

[0018]

[0031] Another aspect disclosed herein is a holographic data storage system that uses one or more waveguide networks to spatially multiplex (i.e., read from / drive to different subvolumes of the medium) across a holographic recording medium without requiring any relative mechanical motion between the medium and the waveguide network. Examples of such systems are described below and utilize active and / or passive light pipes. In the examples described, multimode waveguides can be used to carry an entire digital image to / from the holographic recording medium simultaneously or to carry a reference beam at one of a plurality of possible angles.

[0019] Active light pipe:

[0032] Figures 3A - D show schematic side views of examples of the form of an active light pipe 300 having a particular physical structure. As will be appreciated, this is merely one example of a suitable physical structure that can provide the desired optical configurability. Further examples are considered below.

[0020]

[0033] The active light pipe 300 is shown as having at least a first surface region 300 - 0 and a plurality of active switches in the form of switchable Bragg gratings (SBGs) that can be surface - mounted or volume - embedded. In this example, two such SBGs 300 - 1, 300 - 2 are shown on the first surface 300 - S1 of the waveguide 300, but it will be understood that more SBGs can be placed at appropriate locations on the surface 300 - S1 of the waveguide 300 and / or embedded within the bulk of the waveguide 300. Each SBG 300 - 1, 300 - 2 can be individually controlled so that its reflection / transmission characteristics change in order to transmit or reflect an incident beam. The SBGs 300 - 1, 300 - 2 form respective surface regions of the active light pipe 300, in which regions light can either enter (incouple) into the waveguide 300 or exit (outcouple) from the waveguide 300 depending on how the waveguide 300 is being used.

[0021]

[0034] The first surface region 300 - 0 is an end region of the waveguide 300 from which the first side surface 300 - S1 of the waveguide extends along the axis 301 of the waveguide 300.

[0022]

[0035] Figures 3E and 3F show cross - sectional views of the waveguide 300, which in this example has a rectangular - shaped cross - section, and it can be seen that four side surfaces 300 - S1, 300 - S2, 300 - S2, 300 - S4 extend along the axis 301 of the waveguide 300. In this example, as depicted in the figures, the SBGs 300 - 1, 300 - 2 are all located along the first side surface 300 - S1, but in general, such SBGs can be attached to multiple surfaces of the waveguide 300 depending on the application.

[0023]

[0036] The SBGs 300-1 and 300-2 are positioned along the first side surface 300-S1 of the waveguide such that the distance from the first region 300-0 increases, and the first SBG 300-1 is located closest to the first region 300-0.

[0024]

[0037] FIGS. 3A, 3B, and 3E depict a "one-to-many" use case, where the first surface region 300-0 serves as the in-coupling region, and the second surface regions 300-1 and 300-2 of the SBGs serve as the out-coupling regions. As an example, FIG. 3 shows a first light ray 304 being introduced into the waveguide 300 through the in-coupling region 300-0. In this example, the first surface region 300-0 is angled with respect to the side surfaces 300-S1,..., 300-S4, such that the first light ray 304 enters the bulk of the waveguide 300 through the first surface region 300-0 at an angle sufficient to achieve total internal reflection at each of the side surfaces 300-1,..., 300-4 within the waveguide 300.

[0025]

[0038] Each of the SBGs 300-1 and 300-2 can be configured to change between a reflective state and a transmissive state. FIG. 3A shows a configuration where the first SBG 300-1 is in the reflective state, and the incident light ray 300 is reflected by the first SBG 300-1 and guided along the waveguide 300 until it reaches the second SBG 300-2. The SBG 300-2 is shown in the transmissive state, and the light ray 304 diffracts out of the waveguide 300 through the second SBG 300-2, thereby being extracted from the waveguide 300 through the surface region of the second SBG 300-2. This configuration of the SBGs 300-1 and 300-2 creates a "channel" through the waveguide 300 between the first surface region 300-0 and the surface region of the second SBG 300-2.

[0026]

[0039] In contrast, FIG. 3B shows that the first SBG 300-1 is in a transmissive state. Accordingly, as soon as the first light ray 304 reaches the first SBG 300-1, it diffracts out of the waveguide 300 through the first SBG 300-1 instead, and is thereby taken out of the waveguide 300 through the surface region of the first SBG 300-1 instead. With this configuration, a channel through the waveguide 300 is created between the first surface region 300-0 and the surface region of the first SBG 300-1.

[0027]

[0040] In this way, it is possible to guide the first light ray 304 through the waveguide 300 from the first region 300-0 until it exits the waveguide 300 at the surface region of the SBG 300-1 or 300-2. For the sake of simplicity, only two SBGs 300-1, 300-2 are described in relation to each other, but it will be understood that the same principle can be applied to more SBGs.

[0028]

[0041] FIG. 3E shows, in cross-section, how the first light ray 304 can propagate from some or all of the side surfaces 300-S1, ..., 300-S4 depending on the angle of the first light ray 304 via TIR.

[0029]

[0042] As depicted in FIGS. 3C, 3D and 3F, it is equally feasible to use the described active light pipe 300 for many-to-one optical transmission.

[0030]

[0043] Figure 3C shows the same SBG configuration as in Figure 3A. The only difference is in the way the waveguide 300 is used, where a second light ray 308 incident on the second SBG 300-2 from an external source (not shown) is depicted. Due to the second SBG being in a transmissive state, the second light ray 308 diffracts into the waveguide 300 through the second SBG (providing incoupling in its surface region there), and is then guided from there through the waveguide 300 to the first surface region 300-0 (here, the outcoupling region). This includes reflection from the first SBG 300-1 which is in a reflective state here. The reflective state of the first SBG 300-1 prevents the light ray 308 from exiting the waveguide through the first SBG 300-1. Moreover, an external light ray 309 that might accidentally be incident on the first SBG 300-1 is essentially reflected away from the first SBG 300-1 and thus does not enter the waveguide 300.

[0031]

[0044] Figure 3D shows the same configuration as in Figure 3B, but here the second light ray 308 is incident on the first SBG 300-1 from an external source. Due to the first SBG 300-1 being in a transmissive state, a third light ray 310 enters the waveguide 300 at that location by diffraction and is guided to the first surface region 300-0.

[0032]

[0045] Figure 3F shows how the second light ray 308 can propagate within the waveguide 300 in cross-section. The same explanation as in Figure 3E applies, but the direction of the light ray is reversed.

[0033]

[0046] The above description assumes perfect reflectivity / transmittance of the SBGs in the transmit / reflection states. As is understood, this is not actually an absolute condition, and more generally, the system has a certain tolerance for imperfections in the SBGs 300-1, 300-2 and the waveguide 300. Appropriate signal processing techniques for compensating for the distortion introduced into the waveguide 300 will be described later.

[0034]

[0047] Although SBG 300-1 and 300-2 are depicted as separate elements, in fact, separate independently controllable regions of a single large SBG can extend over all or most of the first side 300-S1.

[0035]

[0048] SBG is just one possible form of active switching element. For example, in a polarized beam, the same effect can be achieved using a controllable polarization filter attached to the surface of the waveguide 300 or embedded within the bulk of the waveguide. SBG and the controllable polarization filter are examples of non-mechanical active switches and can change the optical properties of the waveguide 300 via non-mechanical effects. Other examples of guiding elements include controllable mirrors such as micro-mirror devices or other micro-electro-mechanical systems (MEMS), which are examples of mechanically induced elements.

[0036]

[0049] When using a polarization filter as the guiding element, SBG 300-1 and 300-2 can be replaced by passive diffraction elements, and the polarization filter operates to controllably direct the beam to or away from the passive diffraction element as needed without the need to reconfigure the diffraction element.

[0037]

[0050] Note that even if those guiding elements themselves are mechanical, the need for mechanical operation of the entire waveguide 300 is still avoided.

[0038] Active Light Pipe Network

[0051] As used herein, a "waveguide network" can take the form of a single waveguide or a network of multiple interconnected waveguides. A waveguide network comprising a plurality of active light pipes has certain advantages in terms of flexible optical data transmission.

[0039]

[0052] FIGS. 4A and 4B show a side view of one side of a waveguide network (a part thereof) including first and second active light pipes 400, 420. The second light pipe 420 has a first surface region 420-0 located adjacent to a corresponding surface region of the first light pipe 400 to receive a beam from the second waveguide 420 or direct a beam to the second waveguide 420 through the first surface region 400-0. By way of pure example, a light ray 404 is shown propagating through the first waveguide 400 to a corresponding surface region of the first waveguide 400, and the corresponding surface region of the first waveguide 400 exists adjacent to the first surface region 420-0 of the second waveguide 420. The light ray 404 is taken out from the first waveguide 400 through an SBG 400-1 attached to an adjacent surface region of the first waveguide 400 and taken into the second waveguide 420 through the first surface region 400-0. From there, the light ray 404 can be directed to any one of a plurality of SBGs 420-1, 420-2 of the second waveguide 420 in a one-to-many manner. In a many-to-one manner, the same arrangement can be used to direct a beam in the other direction from the second waveguide 420 to the first waveguide 400 by reversing the direction of the light ray.

[0040]

[0053] This example considers two mutually coupled waveguides 400, 402, but the principle can be applied to more mutually coupled waveguides to enable flexible data routing through the waveguide network.

[0041]

[0054] More generally, the surface region of the medium can be optically coupled to the corresponding surface region of the waveguide in another way, for example, through an air interface or one or more other optical components (which may themselves be waveguides and may or may not provide active or passive switching functionality).

[0042] Holographic storage device

[0055] Here, the application of the active light pipe to a holographic storage device will be described.

[0043]

[0056] FIGS. 1A and 1B show a schematic perspective view of a holographic recording medium 102, which is a volume of a relatively thick photosensitive material that can persistently store an optical pattern as a “hologram” embodied within the holographic recording medium 102 (which can simply be referred to as medium 102 for simplicity). The hologram is generated by exposing that sub-volume 110 (region) of the medium 102 to an optical pattern such that a persistent state change occurs within the sub-volume 110. The hologram generated in the sub-volume 110 by that state change records the optical pattern to the medium 102 and can reproduce the optical pattern therefrom at a later time. The hologram is persistent in that the medium 102 does not require power to maintain it once it is generated. The composition and structure of the medium 102 can be such that the hologram cannot be erased once it is generated (thus providing the form of a “write once read many” (WORM) storage device) or such that the hologram can be erased and replaced (however, it persists until and unless it is erased).

[0044]

[0057] A single hologram can record an optical pattern encoding a very large number (e.g., millions) of bits, thereby enabling very large amounts of data to be written to / read from the holographic recording medium 102 in parallel (simultaneously) through the exposure / regeneration of the optical pattern. Another advantage of the holographic storage device is that many holograms can be written to the same sub-volume 110 of the holographic recording medium 102, thereby significantly increasing the data storage capacity per unit volume of the holographic recording medium 102.

[0045]

[0058] More specifically, FIG. 1A shows how input beam 104 and reference beam 106 are directed at subvolume 110 through first and second sides 102-4, 102-6 of medium 102, respectively, to write a set of data to medium 102. Thereby, an optical pattern in the form of an interference pattern resulting from the interference between input beam 104 and reference beam 106 is generated. If beams 104, 106 have sufficient power and subvolume 110 is exposed for a sufficient duration, the interference pattern generated by interfering beams 104, 106 will be permanently recorded as a hologram within subvolume 110. As will be explained below, the set of data is embedded in input beam 104 and can be retrieved from the resulting hologram. Thus, the set of encoded data is written to subvolume 110. In the following example, the set of data is encoded as a digital image and then embedded in input beam 104 via spatial modulation.

[0046]

[0059] As shown in FIG. 1B, to read data from subvolume 110, a collimated reference beam 116 is directed at subvolume 110 through second side 102-6 of medium 102, and collimated reference beam 116 interacts with the hologram to generate an output beam 108, which essentially matches input beam 104 used to write the hologram to the extent that the embedded data can be retrieved from output beam 110. Output beam 108 propagates out of subvolume 110 through third side 102-8 of medium 102.

[0047]

[0060] The reference beam 116 used to read the data is substantially aligned with the reference beam 106 originally used to write the data and is directed at an angle (or more generally a direction) that is precisely aligned with the angle of the original reference beam 106. This is because the ability to read the hologram (i.e., generate the output beam 108 from which the data can be restored) is highly sensitive to the angular deviation between the reference beam 106 used for writing the hologram and the reference beam 116 used for reading the hologram. This sensitivity can be utilized to record multiple holograms within the same sub-volume 110, each hologram being generated using a different reference beam angle, and two completely different holograms can be generated using only a very small difference in the reference beam angles. Thus, many (e.g., hundreds or thousands) of holograms can be written into the same sub-volume 110, each encoding many (e.g., millions) of bits.

[0048]

[0061] Figure 2A shows a schematic perspective view of an example of a holographic memory system 200 incorporating certain principles of the present disclosure. In this particular example, three separate waveguides 204, 206, 208 are used to carry the input beam 104, reference beams 116, 126, and output beam 118, and the three separate waveguides 204, 206, 208 can be individually referred to as an input waveguide 204, a reference waveguide 206, and an output waveguide 208. As described, the terms "optical waveguide" and "light pipe" are used interchangeably herein. Each of the waveguides 204, 206, 208 provides spatial multiplexing in the sense that it can direct a signal to or from any one of a plurality of sub-volumes within the holographic recording medium 102 (in the case of the input and reference waveguides 204, 206) or from any one of a plurality of sub-volumes within the holographic recording medium 102 (in the case of the output waveguide 208). This provides spatial multiplexing across the volume of the holographic recording medium 102 without the need for any mechanical movement of the waveguides 204, 206, 208 relative to the holographic recording medium 102. To avoid the need for such mechanical movement, the guiding elements are located on or within each of the waveguides 204, 206, 208 and are configured such that the optical properties of the waveguides 204, 206, 208 change to direct a signal to or from different sub-volumes of the medium 102. That is, to create different channels within the waveguides 204, 206, 208 as needed. In this particular example, the guiding elements take the form of active optical switching elements (switches). The forms that the switches can take are varied. In this example, the switches take the form of SBGs located in different surface regions of the waveguides 204, 206, 208 in the same general arrangement as FIGS. 3A - E. That is, each of the waveguides 204, 206, 208 takes the form of an active light pipe, and each of the waveguides 204, 206, 208 has the same general physical structure as the active light pipe 300 of FIGS. 3A - E.

[0049]

[0062] Each of the waveguides 204, 206, 208 is arranged such that its first surface (i.e., the surface on which its SBG is located) is adjacent to a different side surface of the medium 102, whereby the SBG extends along that side surface of the medium 102. The first and second SBGs of each of the waveguides 204, 206, 208 are respectively denoted by reference numerals 204-1, 204-2, 206-1, 206-2, 208-1, 208-2, and they can all be configured in the manner described above. Additional SBGs are depicted without reference numerals, and the number of SBGs can be selected to adapt to holographic recording media 102 of any size. The following description will, for the sake of brevity, refer to the first and second SBGs of each of the waveguides 204, 206, 208, but it will be understood that the explanation also applies to more SBGs.

[0050]

[0063] Figures 2B - 2D show how the input and reference waveguides 204, 206 are used to write data to the medium 102 in a one-to-many fashion. Figure 2B shows a schematic plan view of the system 200, and Figures 2C and 2D show side views of one side on which the input and reference waveguides 204, 206 can be seen respectively. The input waveguide 404 is used to direct the input beam 104 to any one of a plurality of sub-volumes of the medium 102 through any one of the SBGs 404-1, 404-2 of the input waveguide 204 in the manner described above. The reference waveguide 406 is configured to simultaneously direct the reference beam 106 to the same sub-volume, generating the desired interference pattern to be written to that sub-volume. In the example depicted, both the input waveguide 204 and the reference waveguide 206 are here configured to direct the input and reference beams 104, 106 respectively to the sub-volume denoted by reference numeral 110 through the second SBGs 204-2, 206-2 of each of the waveguides 204, 206 respectively.

[0051]

[0064] Figures 2E through 2G illustrate how reference and output waveguides 206, 208 can be used to read data from medium 102. FIG. 2E is a plan view, and FIGS. 2F and 2G are side views of one side from which reference and output waveguides 204, 206 are visible. The reference waveguide 206 is used in exactly the same manner as depicted in FIGS. 2B through 2D, except that here, the reference beam 116 is directed to the subvolume (subvolume 110 in this case) from which the hologram is read. The output waveguide 208 is used in a one-to-many fashion to direct the resulting output beam 108 from subvolume 110 through waveguide 208 for subsequent detection.

[0052]

[0065] Each subvolume 110 may have a height and width of several millimeters, for example, as measured along any side, which would generally be sufficient to store millions of pixels per data “page” (e.g., multiplexing angle). In this case, the subvolume volume would be sufficient to store (millions of pixels) * (number of multiplexing angles).

[0053]

[0066] The guiding elements (SBG in this example) of input waveguide 204 and reference waveguide 206 are configured to provide channels for input beam 104 and reference beams 106, 116 from a beam source (emission system) to the subvolume 108 being read, as needed. In SBG, this involves setting the SBG to a transmission or reflection state as needed to create the channels. Similarly, the guiding element (also SBG in this example) of output waveguide 208 is similarly configured to provide a channel from the subvolume 108 being read to the detector. For additional context, this will be described in more detail below with reference to the multi-waveguide network depicted in FIG. 5. However, the principles described in connection with the specific example of FIG. 5 apply more generally to other waveguide network topologies, such as both simpler networks (e.g., a single waveguide) and more complex waveguide networks.

[0054]

[0067] As described above, this enables spatial multiplexing across the medium 102 without mechanical movement of the medium 102 with respect to the waveguides 204, 206, 208. This holds true regardless of the form taken by the guiding elements (as described above, those guiding elements themselves can be mechanical or non-mechanical).

[0055] Holographic memory device using a multi-waveguide network

[0068] FIG. 5 shows an example of a holographic memory system incorporating a multi-waveguide network of the type shown in FIG. 4.

[0056]

[0069] The input waveguide network is shown to include a first input light pipe 203 ("parent" waveguide) to which a plurality of second input light pipes 204A, 204B ("child" waveguides) are coupled. The input beam 104 from the emitter system 504 is taken into the first input waveguide 203 through its incoupling region and can be directed from there to either the second input waveguide 204A or 204B.

[0057]

[0070] The reference waveguide network is shown to include a first reference waveguide 205 to which a plurality of second reference waveguides 206A, 206B are coupled. The reference beams 106, 116 from the emitter system 504 are similarly taken into the first reference waveguide 205 and can be directed to either of the second reference waveguides 206A, 206B.

[0058]

[0071] The output waveguide network is shown to include a first output waveguide 207 to which a plurality of second output waveguides 208A, 208B are coupled.

[0059]

[0072] With the described arrangement, beams can be directed to / diverted from different subvolumes of the plurality of pieces 102A, 102B of the holographic memory medium.

[0060]

[0073] FIG. 4 shows the input beam 104, reference beams 106, 116, and output beam 108, and it will be understood that the sub-volumes will typically be written and read at different times in the manner described above with reference to FIGS. 2A - G.

[0061]

[0074] The first group 204A, 204B, 204C of second waveguides (one each of input, reference, and output) are located around the first piece 102A (first medium) of the holographic storage medium, and the second group 204B, 206B, 208B of second waveguides are located around the second piece 102B (second medium), each having the same general arrangement as FIGS. 2A - G. Thus, the input and reference beams 104, 106, 116 can be directed to any sub-volume of any medium piece 102A, 102B by first guiding those beams to the desired second waveguide of the input and reference networks respectively, and then to the desired sub-volume of the medium piece adjacent to the desired waveguide.

[0062]

[0075] The output waveguide network can be used to direct the output beam 108 from any sub-volume of any of the media pieces 102A, 102B to the first output waveguide 207 from the applicable second output waveguides 208A, 208B, and then from the first output waveguide 207 through the out-coupling region of the first output waveguide 207 detector 508 to the detector. To perform a read from a particular sub-volume, the SBG is configured to provide a channel from that sub-volume to the detector. Thus, in this case, SBGs 204A-2 and 207A-1 are set to the transmissive state, and the other SBGs of the output waveguide network are set to the reflective state as needed to provide a channel for the output beam 108 to the detector 508 (e.g., in this case, SBG 207-2 of the first output waveguide 207 is set to reflect to prevent the propagation of the output beam 104 to waveguide 207-2). The other SBGs of the output waveguide network can be set to reflect to the extent necessary to prevent the transmission (i.e., "leakage") of unwanted light from other regions of the same media piece 102A or from a different media piece 102B (e.g., in this example, SBG 204A-1, which is close to the sub-volume being read, is shown as being set to reflect to prevent unwanted leakage).

[0063]

[0076] In the example above, three separate waveguide networks are used for the input, reference, and output beams 104, 106, 116, 108, but this is not necessarily required. For example, the same waveguide network can be used to carry both the input beam 104 and the reference beams 106, 116, and / or the same waveguide network can be used to carry the input beam 104 and the output beam 108, and / or the same waveguide network can be used to carry the output beam 108 and the input beam 104. In general, having three separate networks is expected to provide optimal performance, but nevertheless, fully functional implementations exist that use only one or two waveguide networks.

[0064]

[0077] Although not depicted in any of the figures, a fourth waveguide network can be used to carry the beam to the remaining sides of the media pieces 108A, 108B. For example, the fourth network can be used to carry an erase beam, which is at least suitable for erasing the hologram therefrom (in the case of an erasable holographic storage device), to a desired subvolume.

[0065]

[0078] Figure 9 shows an alternative physical structure where a single "slab" of the holographic medium 102 is used instead of the individual pieces 102A, 102B of FIG. 5. An input waveguide network is depicted and has essentially the same physical configuration, but here the second input waveguides 204A, 204B are configured to direct the input beam 104 to different subvolumes of the same slab 102. In FIG. 5, each of the second input waveguides 204A, 204B provides one-dimensional multiplexing along the length of a different single media piece 102A, 102B, while in FIG. 9, the second waveguides 204A, 204B provide two-dimensional spatial multiplexing across the slab of the holographic medium 102 (each waveguide provides one-dimensional multiplexing individually, but there is 2D multiplexing across the slab 102 as a whole).

[0066]

[0079] The system of FIG. 9 is limited to a maximum of two waveguide networks (one on each side of the slab 102). As described above, this is still a viable arrangement since the same network can be used to carry multiple beams.

[0067] Data Encoding

[0080] FIG. 6A shows an example of an emitter system 504 that provides both the input beam 104 and the reference beam 106. The input beam is an expanded spatially modulated laser beam. The laser 600 emits a coherent narrow laser beam, and the coherent narrow laser beam is split using a beam splitter 604.

[0068]

[0081] A portion of the beam from beam splitter 602 is used as reference beam 106. In this example, a controllable reference beam steering element 612 is used to manipulate reference beam 106 to enter reference waveguide 106 at a desired angle. By changing the angle of reference beam 106 before it is incorporated into reference waveguide 206, different holograms can be written to / read from the same sub-volume of the medium in the manner described above.

[0069]

[0082] As an alternative to or in addition to beam angle multiplexing, multiple patterns can be stored in and read from the same sub-volume with different phases (phase multiplexing) of reference beams 106, 116. Thus, a logical address can correspond to a particular reference beam angle and / or phase characteristic. All explanations regarding modulation of the reference beam angle apply equally to phase modulation.

[0070]

[0083] Another portion of the beam from the beam splitter 602 is expanded using the beam expander 604, and the expanded beam passes through the spatial light modulator (SLM) 606. The encoder 610 receives a set of data to be encoded, encodes the set of data as a digital image, and then the digital image is modulated via the SLM 606 and embedded in the expanded beam. The in-coupling optical system (in this case, the Fourier lens 608) positioned such that the plane of the SLM 606 is substantially in the focal plane of the Fourier lens 608 is used to separate the expanded beam into entirely different propagation modes, and in this example, the modes correspond to unique propagation directions, where each mode corresponds to a particular point in the plane of the SLM 606. The different propagation modes are captured in the input waveguide 202 and induced therefrom in the manner described above. Using the in-coupling optical system 608, the data is “angle encoded” in the reference beam in the sense that points within the digital image essentially correspond to unique propagation directions (i.e., unique propagation modes of the input beam 104). This is similar to light rays from a distant object that can be regarded as an infinite point. The angle-encoded input beam 104 of FIG. 6A is an example of a plurality of propagation modes (i.e., components propagating in different directions) of a “multimode” optical signal, and this arrangement provides a form of angular diversity.

[0071]

[0084] Note that the term "multimode" does not necessarily imply the use of such an in-coupling optical system 608, nor does it necessarily mean that all image points need to uniquely correspond to a given propagation direction. That is, multimode does not necessarily imply a one-to-one correspondence between propagation modes and image points / data points. For example, FIG. 6B shows an alternative feasible emitter system in which a spatially modulated beam is directly captured into the input waveguide 204. In this case, there are still multiple modes (i.e., multiple spatial paths through the waveguide for any given channel), but there is no one-to-one correspondence between the propagation direction and the image points, and in some cases, there is also no one-to-one correspondence between the image / data points and the modes. This provides a form of spatial diversity based on the form of MIMO (multiple input multiple output) transmission via a plurality of pixels of the SLM 606 and a detector array of the spatial coherent detector 508.

[0072] Data decoding

[0085] FIG. 7 shows a spatial coherent detector 508 used to measure the optical field of the output beam 108. In contrast to conventional "direct detection", the spatial coherent detector 508 includes an array of pixels (or more generally detector elements), each of which is configured to measure both the amplitude and phase (not just the intensity) of the optical field at the location of that pixel. These can be measured, for example, using a local oscillator 712 of the spatial coherent detector 508. Thus, the array of pixels can measure the phase and amplitude variations of the optical field in both time and space, and thus can provide an analog or digital representation of the measured optical field. In this case, the optical field to be measured is the optical field of the output beam 108 guided to the spatial coherent detector 208 via the output waveguide 208.

[0073]

[0086] Only a single array is depicted, but in fact, there can be multiple physical arrays that cooperate as a single "logical array". For example, this logical array can be split across two physical cameras.

[0074]

[0087] The physical detector array can take the form of a single camera (where each detector element is a pixel or set of pixels of the camera), or can take the form of multiple cameras. In an extreme case, each detector element can be a separate camera, in which case the logical detector array can potentially be split across a very large number of physical detectors.

[0075]

[0088] As described, the route from a particular incoupling region where the beam enters the waveguide network to a particular outcoupling region where the beam exits the waveguide network (these regions may be in the same or different waveguides) can be referred to herein as a "channel". As described, in a multimode waveguide network, a single channel encompasses multiple spatial paths. The output beam 108 will be guided through a particular channel of the output waveguide network (i.e., from its particular incoupling region to the outcoupling region of the output waveguide 208). Moreover, the output beam 108 will be generated from a hologram generated using the input beam that was guided from the incoupling region of the input waveguide network to its particular outcoupling region. The hologram will be generated and read using a reference beam that was similarly guided through a particular channel through the reference waveguide network. All of the input beam 104, reference beams 106, 116, and output beam 108 are susceptible to the effects of distortion within the associated waveguide network that is specific to the channel through which they are guided. To compensate for such distortion, the signal processing component 700 applies analog and / or digital signal processing to the representation of the measurement field. The signal processing component 700 does this using a channel model associated with the subvolume that is currently being read (i.e., where the output beam 108 was generated). The channel model associated with a particular subvolume models not only the channel through which the output beam 108 is guided to the detector 508, but also the channels through which the input beam 104 used for writing the hologram is guided to that subvolume and the channels through which the reference beams 106, 116 used for writing / reading to the hologram are guided to that subvolume.

[0076]

[0089] Each channel model can take the form of, for example, a transfer function (modeling the channel directly) or an inverse transfer function (modeling the channel with its approximate inverse function). Note that the transfer function is applied to the representation of the measured optical field (i.e., both its measured phase and amplitude at different spatial points), not just to the intensity of the light. Spatial coherent detection provides a greater scope for removing or reducing such channel distortions, the aim being to restore the original digital image accurately enough so that the decoder 704 can easily decode the encoded data from the restored image.

[0077]

[0090] Signal processing 700 can correct phase coherence and noise, for example, using a combination of optical techniques and computational techniques (which may include, for example, machine learning techniques).

[0078]

[0091] Although described in the context of a holographic memory device, the use of such signal processing 700 in combination with spatial coherent detection is not limited in this regard and can be applied in other contexts such as optical communication or optical computing, or any other context where the received output beam is susceptible to distortions introduced in one or more waveguide networks.

[0079]

[0092] FIG. 7 shows an out-coupling optical system 715 arranged to essentially reverse the effect of the in-coupling optical system 608 of FIG. 6B (i.e., essentially resolve each propagation mode to a single point in the plane of the spatial coherent detector 508). Again, this is not essential, and the out-coupling optical system 715 can be omitted by using the alternative emission system of FIG. 6B.

[0080]

[0093] Although not depicted in FIGS. 6A or 6B, a certain level of preprocessing can be applied to the digital image before modulating and embedding it into the input beam 104. This can reduce the required compensation level on the detector side. Even if such preprocessing is performed, some level of detector-side processing can be applied to account for different distortion effects between different channels.

[0081] Dynamic Scheduling

[0094] FIG. 8A shows a controller in the form of a scheduler 800 that can schedule read and write operations within a holographic memory system of the type described above. To facilitate effective scheduling, sub-volumes within the medium 108 or within each medium piece 108A, 108B are assigned unique addresses. This provides a form of an addressable holographic memory device similar to more conventional forms of addressable electronic memory devices. However, there are many features that go beyond conventional addressing.

[0082]

[0095] First, as described above, a single subvolume can store multiple holograms at different reference beam angles. To accommodate this, each address uniquely corresponds to a specific subvolume in combination with a specific reference beam direction (i.e., a unique address is assigned to each available tuple), indicating a specific subvolume within medium 102 or within either one of medium pieces 102A, 102B, and indicating a specific reference beam direction (e.g., an angle or a combination of angles that define the beam direction, where the term "angle" is used as a concise expression for referring to the direction of the reference beam, but it will be understood that the direction can in fact be defined by multiple angles depending on the system configuration). Thus, a subvolume can be associated with potentially many addresses corresponding to different reference beam angles. A tuple defines a logical storage location, and multiple logical storage locations are provided at the physical level by the same subvolume at different reference beam angles. Each of the logical storage locations has a unique address (ADDR). This notation is used as a concise expression for meaning the address corresponding to the subvolume and the reference beam angle, but it will be understood that this does not imply a particular address representation. Any address space and addressing mechanism that uniquely identifies the logical storage locations of this nature can be used.

[0083]

[0096] Second, in contrast to conventional storage devices, each of the logical storage locations can store the entire image, and thus a single logical storage location can potentially store a very large number (e.g., thousands or millions) of bits.

[0084]

[0097] Scheduler 800 operates at the logical storage level and schedules incoming read and write operations for different addresses within an appropriate time interval.

[0085]

[0098] Reference numerals 804, 806, and 808 are each used to denote an input, a reference, and an output optical waveguide network. As described above, each can be a single waveguide network or a multiplexed waveguide network (e.g., similar to FIG. 5), and has one or more configurable guiding elements (e.g., SBG or other active switching elements) that can be used to create channels to different subvolumes of one piece (or pieces) of a holographic storage medium.

[0086]

[0099] During a period (write period) in which a write operation for a particular address is scheduled, the guiding elements in the input and reference waveguide networks 804, 806 are set to create channels for the input beam 104 and the reference beam 106 from the emitter system 504 through their respective input and reference networks 804, 806 to the corresponding subvolumes. In addition, the reference beam steering element 612 is set to direct the reference beam 106 in a corresponding direction into the reference network 806. Thereby, a desired interference pattern is generated at the reference beam angle within the subvolume, and thus, if the subvolume is exposed to the interference pattern for a sufficiently long duration, the interference pattern is persistently stored as a hologram.

[0087]

[0100] During a period (read period) in which a read operation for a particular address is scheduled, the guiding elements in the reference and output networks 806, 808 are likewise set to create a channel for the reference beam 116 from the reference network 806 to the subvolume, and a channel for the output beam 108 from the subvolume through the output network 808 to the detector 508. The reference beam steering element 612 is likewise set to direct the reference beam 116 in a corresponding direction into the reference network 806 to read the intended hologram at the reference beam angle within the subvolume.

[0088] Optimal Scheduling

[0101] One problem with holographic memory devices is that they require a relatively long time to complete a write operation. This is because in order to generate a stable (persistent) hologram, it is necessary to expose the sub-volume for a relatively long time interval compared to the time required to read an existing hologram. For example, in some systems, an exposure time of about 10 or 100 milliseconds may be required to continuously generate a hologram. In contrast, the read operation only needs to expose the sub-volume for the time required to measure the optical field resulting from output beam 108. For example, in some systems, an exposure time for the read operation (i.e., the duration of the medium exposure and / or the shutter opening time of the camera for reading the beam) of about 100 microseconds to 1 millisecond can be achieved, and the write operation can be segmented, for example, over a similar duration interval.

[0089]

[0102] However, it is recognized herein that it is not necessary to perform the write operation in a single continuous time interval. As long as the sub-volume is exposed to the interference pattern over a sufficient total time, the exposure can be performed "stepwise". To perform the write operation, the sub-volume can be exposed to the interference pattern to be recorded for any length of time until it continuously records that interference pattern. Then, the exposure can be terminated to perform some other operation and resumed at a later time. The intervals between write intervals can generally be arbitrarily long, and since the hologram data cannot be read from the medium until the complete exposure is finished, it is assumed that the data can be held elsewhere (e.g., in a non-volatile buffer) during that time. However, this can result in a significant write latency. Therefore, the scheduler 800 is required to balance the amount of such buffering required by the need to avoid head-of-line blocking during reading.

[0090]

[0103] FIG. 8B schematically shows a scheduling scheme that exploits this effect to efficiently schedule a set of read and write operations. For purposes of illustration, a single write operation 814 and three read operations 818a, 818b, 818c are depicted, but the scheduling scheme can be applied to any number of read and write operations.

[0091]

[0104] Similar to energy efficiency, an additional advantage in staggering write operations is that read operations can be scheduled in a way that reduces "head of line blocking," which otherwise could have a "naive" scheduling scheme, which simply schedules write operations in a time interval long enough until completion and performs read operations when the write operation is completed. The "naive" scheduling scheme is the way write operations are typically scheduled in conventional memory systems. In such a system, write operations have the effect of significantly increasing the latency of read operations 818a - c such that read operations have to wait until write operation 814 is completed.

[0092]

[0105] By performing write operation 814 step - by - step over multiple discontinuous time intervals and scheduling read operations 181a - c in read intervals that are temporally interleaved with the multiple write intervals, the dual effect of increased energy efficiency and reduced read operation latency is achieved.

[0093]

[0106] In each write interval, the sub - volume involved in write operation 814 is exposed to the same interference pattern that encodes the image to be stored. Input and reference networks 804, 806 are configured to provide the necessary channels in the same way in each write interval, and reference beam 106 is directed in the same direction.

[0094]

[0107] During the intermediate read interval, the reference and output networks 806, 808 can be reconfigured, and the reference beam 116 can be redirected as needed to read any desired hologram within the holographic memory system.

[0095] Energy optimization

[0108] Surprisingly, when the write operation is segmented in the manner described above, the total time required to continuously generate the hologram (i.e., the sum of the individual durations of the plurality of discontinuous write intervals) is found to be less than the exposure time that would be required to write the hologram in a single continuous write interval for a given output of the input and reference beams 104, 106. For example, a write operation that may require approximately 10 ms to complete in a single continuous write interval has been found to be achievable over a plurality of discontinuous time intervals with a total duration of only approximately 7 ms in the same system. This in turn means that by segmenting the write operation in the manner of FIG. 8B, the amount of energy required to complete the write operation (which is essentially directly proportional to the total duration required to complete the write operation with a beam of fixed output) can be reduced.

[0096]

[0109] FIG. 13 shows a set of results demonstrating this effect for a particular one setting. The results for pulsed and continuous writing are indicated by reference numerals 1302 and 1304, respectively. In this particular example, for each write time exceeding approximately 30 ms, it can be seen that a higher diffraction efficiency is achieved with pulsed writing having that amount of total duration than with continuous writing having the same total duration. It can be seen that the diffraction efficiency increases with an increase in the write time. In this regard, the diffraction efficiency indicates the total energy diffracted within the medium 102, which is proportional to the incident energy. Thus, it can be seen that better energy efficiency is achieved with pulsed writing compared to continuous writing.

[0097]

[0110] Figure 8A shows an energy estimation function 804 that can be used to estimate the amount of energy required to perform a set of operations according to a specific schedule, taking into account the above-described effects. The scheduler 800 attempts to determine a substantially optimal schedule that substantially minimizes the total estimated energy required to complete a given set of operations. This goal is fully consistent with the goal of reducing the impact of head-of-line blocking in write operations. That is, by staging the write operation 814 in the manner described above, not only is the intervening time interval freed up so that the read operations 818a - c can be completed with lower latency in the intervening read intervals, but also the total amount of energy required to complete the write operation 814 is reduced for the reasons discussed in the previous paragraph.

[0098]

[0111] Scheduling can be dynamic in the sense that it can dynamically interrupt the write operation 814 in response to incoming read operations 818a, 818b, 818c that occur or are received after the write operation 814 has started but before it is completed. This ability to dynamically interrupt the write operation 814 allows the scheduler 800 to respond to incoming read operations with minimal latency.

[0099] Cost - function - based scheduling

[0112] Here, one possible scheduling scheme based on cost - function optimization will be described. In this case, the scheduler 800 is tasked with optimizing a defined cost function, and this cost - function optimization involves balancing competing factors such as write latency, read latency (e.g., expressed in terms of read - buffer time), and energy (taking into account the energy - reduction effect from staging the write operation).

[0100]

[0113] For example, a set of read operations R = {r i}, a set of write operations W = {w jGiven a schedule S that schedules these operations in a particular way (i.e., assigns each operation to one or more time intervals that may include discontinuous time intervals for write operations), the cost function is [Number] which can be defined as, where in the formula · The first term L S (r i ) is the estimated latency for the read operation r i when the schedule S is given. More generally, the first term can be any read latency penalty. · The second term L S (w j ) is the estimated latency for the write operation w j when the schedule S is given. More generally, the second term can be any write latency penalty. · The third term E S (w j ) is the estimated energy required to complete the write operation w j when the schedule S is given, and the energy estimation function E S takes into account the energy reduction effect from "staggering" the write operations. More generally, the third term can be any energy penalty (i.e., something that penalizes an increase in the required energy). Here, E S represents the energy estimation function 804. · W1, W2, W3 are weights. For example, the first term may have a higher weight to impose a greater penalty on the read latency than the write latency.

[0101]

[0114] This cost function considers three competing factors, but any number of cost terms (e.g., any two of the above and / or other cost terms) may exist.

[0102]

[0115] Using the second and third terms, staggering the write operations can reduce the associated energy penalty of the third term, but the latency penalty of the second term may increase. The optimizer will attempt to find the optimal schedule S * that balances these competing requirements. S * = argmin S C(S)

[0103]

[0116] That is, the schedule S * substantially minimizes the cost.

[0104]

[0117] Although the above uses the term "cost", this is merely a convenient label. Optimizing a reward function (e.g., one that rewards latency and / or energy reduction) is synonymous in this specification with optimizing a cost function.

[0105] Alternative waveguide network:

[0118] FIGS. 10A and 10B show an alternative system in which spatial multiplexing is achieved by modulating one or more optical properties of the input and reference beams 104, 106, 116 instead. In such a system, passive (non-switchable) guiding elements can be used instead of the active (switchable) guiding elements of the previous figures.

[0106]

[0119] The example of FIG. 10A contemplates frequency (or equivalently wavelength) modulation. In this case, the light pipes themselves are passive and have static wavelength-dependent outcoupling (such as a continuous longer-pass dichroic interference filter or a variable center wavelength bandpass filter).

[0107]

[0120] Figure 11A shows a light pipe 1100 having an outer surface 1100-S, along which there are a plurality of passive filters 1100-1, 1100-2. The configuration of the light pipe 1100 is the same as that in FIGS. 3A - D, except for the fact that the filters 1100-1, 1100-2 replace SGB 300-1, 300-2. The filters 1100-1, 1100-2 have different frequency responses (i.e., they operate as frequency filters). More specifically, each of the filters 1100-1, 1100-2 essentially transmits a relatively narrow range of optical frequencies and essentially reflects frequencies outside that range. FIG. 11A shows an in-coupling beam 1104, whose frequency is within the range of the second filter 1100-2 but outside the range of the first filter 1100-1. Thus, the beam 1104 is reflected by the first filter 1100-1 but transmitted through the second filter 1100-2 (and thus exits the light pipe 1100 at that location). FIG. 11B shows a beam 1104' of a different frequency, where its frequency is within the range of the first filter 1100-1 and thus is transmitted through the first filter 1100-1.

[0108]

[0121] Such a light pipe 1100 can be used instead of the active light pipe described above, and the above description also applies in the same way to the following modifications of the system.

[0109]

[0122] FIG. 10A shows a scheduler 800 communicatively coupled to a laser 600 of an emitter system for varying the frequencies (or equivalently wavelengths) of the input and reference beams 104, 106, 116. In this case, any of the beams can be directed to a desired holographic storage area by correspondingly setting the frequency. Here, different beam frequencies correspond to different routes through the waveguide network (defined by the different frequency characteristics of the passive filters), and the frequencies can be set to correspond to any desired route.

[0110]

[0123] In this case, wavelength is used as the switching dimension. The laser 600 is a fast tunable laser that functions as an active element.

[0111]

[0124] In such a recognition, the switching can be in only one spatial dimension (i.e., along a single pipe). However, using a laser with a sufficient range and narrow linewidth, the first light pipe can filter coarsely (i.e., over a relatively wide wavelength range), and the subsequent light pipes can sample more finely (i.e., over a narrower wavelength range). Another factor limiting the linewidth is the requirement for a relatively long coherence length, and thus, in any event, the linewidth can be made sufficiently narrow. In connection with a holographic memory device, to achieve the replication of the input field to addressable locations over a 2D output space, this implementation can be combined with, for example, a second implementation using different switchable parameters (e.g., polarization).

[0112]

[0125] In connection with the read operation, the frequency of the output beam 108 is matched to the frequency of the reference beam 116 used to read a particular subvolume, and by applying the same principle, an appropriate filter can be used in the output waveguide network 808 to return it to the detector.

[0113]

[0126] FIG. 10B shows an example of such an implementation with a controllable polarization element 601 that can be used to vary the polarization of the input and reference beams 104, 106, 116. This can be combined with a passive polarization filter on or within the light pipe. This passive polarization filter can be implemented as an alternative to or in addition to the passive frequency filter of the example of FIG. 10A. Such polarization modulation provides two independent routes and can be utilized in combination with, for example, passive wavelength filtering and / or an active light pipe. The polarization modulation of the beam can also be combined with an active polarization filter.

[0114]

[0127] Note that all of the various "passive" and "active" implementations described above can be implemented separately or in combination (e.g., a combination of active and passive guiding elements can be used). That is, a waveguide can have both passive and active elements, and / or it is possible to combine active and passive waveguides within the same network.

[0115] Additional hierarchical levels:

[0128] The above examples consider a waveguide network having two hierarchical "levels" of a parent waveguide and a child waveguide. However, a multi-waveguide network can have three (parent, child, grandchild) or more levels. Note that the terms "child", "parent", and "grandparent" do not necessarily imply a direct hierarchical relationship. That is, the term child or grandchild can refer to any waveguide at any hierarchical level below a respective parent or child waveguide. That is, a child / grandchild waveguide can be optically coupled to a parent / child waveguide not only via, for example, an air interface (direct grandchild), but also via one or more of its other child / grandchild waveguides (indirect grandchild).

[0116]

[0129] FIG. 12 shows an example of a waveguide network having three hierarchical levels. The parent waveguide 1200 has two direct child waveguide networks 1202A, 1202B optically coupled to the parent waveguide 1200 in the manner described above, and each of those child waveguides 1202A / 1202B has two grandchild networks 1204A-A, 1204A-B / 1204B-A, 1204B-B optically coupled to the child waveguides 1202A / 1202B in the same manner.

[0117]

[0130] An extreme example is a "binary tree" architecture where all waveguides have exactly two direct children, potentially having more than three levels of waveguides. However, in practice, there can also be situations where it is preferable to increase the number of direct children in order to reduce the number of hierarchical levels required.

[0118]

[0131] More generally, the scheduling techniques described can be implemented in any holographic memory system (with or without waveguides) having the same energy savings advantage. This includes, for example, systems without spatial multiplexing or systems where spatial multiplexing is achieved using mechanical means (such as a movable medium or a movable read / write head).

[0119]

[0132] The schedulers 800 and energy estimation functions 804 shown in FIGS. 8A, 10A, and 10B are functional components of the system. Similarly, the encoder 610, decoder 704, and signal processing components 700 are also functional components. Such components can be implemented in software (i.e., as program code executed on one or more programmable hardware processors such as a CPU, an accelerator, e.g., a GPU), or using other forms of processor hardware such as a field programmable gate array and / or an application specific integrated circuit. The signal processing performed by the signal processing component 700 can be analog or digital signal processing or any combination thereof. Such program code and other data (e.g., the channel model 702) can be encoded in a computer-readable storage device. Examples of computer-readable storage devices include optical, magnetic, and / or solid-state storage devices, which can store code, data, and the like in a non-transitory form. This is in contrast to transitory media such as a transitory signal carrier wave.

[0120]

[0133] A first aspect of the present specification is a method for performing a writing operation in a holographic data storage system, the method including scheduling, by a scheduler, at least one writing operation over a plurality of discontinuous writing intervals, the writing operation being related to a set of data to be stored in an area of a holographic recording medium, and exposing, in each of the discontinuous writing intervals, an area of the holographic recording medium to an interference pattern generated by interference between a reference beam and an input beam carrying the set of data, the plurality of discontinuous writing intervals having a total duration long enough to cause a persistent change in the exposed area, whereby the set of data is recoverable from the area by the end of the last writing interval of the plurality of discontinuous writing intervals.

[0121]

[0134] In an embodiment, the step of scheduling may further include the scheduler scheduling at least one additional operation in at least one interval interleaved with the plurality of discontinuous writing intervals.

[0122]

[0135] The at least one additional operation may include at least one reading operation.

[0123]

[0136] The at least one additional operation may include at least one additional writing operation scheduled over an additional plurality of discontinuous writing intervals interleaved with the plurality of discontinuous writing intervals.

[0124]

[0137] The scheduler can perform the scheduling using an energy estimation function, which is for estimating the amount of energy required to complete at least one writing operation.

[0125]

[0138] The scheduler can perform the scheduling by substantially optimizing a cost function that penalizes an increase in the estimated energy amount required to complete at least one write operation.

[0126]

[0139] The cost function can also penalize an increase in the estimated latency for at least one write operation.

[0127]

[0140] Alternatively or in addition, the cost function can also penalize an increase in the estimated latency for at least one read operation and / or an increase in the estimated latency for at least one additional write operation and an increase in the estimated energy amount required to complete at least one additional write operation.

[0128]

[0141] The scheduler can dynamically schedule at least the later ones of the plurality of discontinuous time intervals in response to a read operation, either between or after the earlier ones of the plurality of discontinuous time intervals.

[0129]

[0142] A second aspect of the present specification is a holographic memory system, including at least one processor configured to implement a scheduler, the scheduler being configured to schedule at least one write operation over a plurality of discontinuous write intervals, the write operation being related to a set of data stored in a region of a holographic recording medium, at least one processor, and an emitter system configured to emit a reference beam and an input beam in which the set of data is embedded in each of the plurality of discontinuous write intervals, and a holographic recording medium, and exposing the region of the holographic recording medium to an interference pattern generated by interference between the input beam and the reference beam in each of the plurality of discontinuous write intervals, thereby providing a holographic memory system arranged to perform the scheduled write operation.

[0130]

[0143] The scheduler may be configured to schedule at least one additional operation in at least one interval interleaved with a plurality of discontinuous write intervals, and the system may further include one or more optical waveguide networks arranged to direct an input beam and a reference beam to regions of a holographic recording medium in each of the plurality of discontinuous time intervals, and the system may use at least one of the one or more optical waveguide networks in the interleaved interval to direct the reference beam to another region of the holographic recording medium or another holographic recording medium, with or without changing the angle and / or phase characteristics of the reference beam, or to the same region of the holographic recording medium but with different angle and / or phase characteristics.

[0131]

[0144] At least one optical waveguide network may have at least one configurable guiding element, and at least one processor may be coupled to the at least one guiding element to reconfigure the at least one guiding element to direct the reference beam to another holographic recording area during the interleaved interval.

[0132]

[0145] At least one optical waveguide network may have at least one static guiding element responsive to controllable optical characteristics of the reference beam, and at least one processor may be coupled to the emitter system to change the optical characteristics of the reference beam to direct the reference beam to another holographic recording area during the interleaved interval.

[0133]

[0146] The write operation and the additional write operation may be associated with respective logical addresses, each logical address defining a holographic storage area for the associated operation and the reference beam angle and / or phase characteristics.

[0134]

[0147] A third aspect of the present specification is a scheduler embodied as executable program code stored in a computer-readable storage medium, which, when executed on at least one processor of a holographic data storage system, is configured to perform any of the above-described methods.

[0135]

[0148] It will be understood that the above-described embodiments are merely illustrative examples. Other variations or use cases of the disclosed techniques will become apparent to those skilled in the art at the time the disclosure of this specification is given. The scope of the present disclosure is not limited by the described embodiments, but only by the appended claims.

Claims

1. 1. A method for performing a write operation in a holographic data storage system, comprising: scheduling, by a scheduler, at least one write operation across a plurality of non-contiguous write intervals, the write operation relating to a set of data to be stored in an area of ​​the holographic recording medium; exposing, in each of the discontinuous writing intervals, the area of ​​the holographic recording medium to an interference pattern produced by interference between a reference beam and an input beam carrying the set of data; wherein the plurality of discontinuous write intervals have a total duration sufficiently long to cause a sustained change of state in the exposure area, such that the set of data is recoverable from the area by the end of a last write interval of the plurality of discontinuous write intervals.

2. The method of claim 1 , wherein the step of scheduling further comprises the scheduler scheduling at least one additional operation in at least one interval interleaved with the plurality of non-contiguous write intervals.

3. The method of claim 2 , wherein the at least one further operation comprises at least one read operation.

4. The method of claim 2 or 3, wherein the at least one further operation comprises at least one further write operation scheduled across a further plurality of non-contiguous write intervals interleaved with the plurality of non-contiguous write intervals.

5. 5. The method of claim 1, wherein the scheduler performs the scheduling using an energy estimation function, the energy estimation function for estimating an amount of energy required to complete the at least one write operation.

6. 6. The method of claim 5, wherein the scheduler performs the scheduling by substantially optimizing a cost function that penalizes increases in the estimated amount of energy required to complete the at least one write operation.

7. The method of claim 6 , wherein the cost function also penalizes an increase in an estimated latency for the at least one write operation.

8. The cost function is: an increase in the estimated latency for the at least one read operation; and / or an increase in the estimated latency for the at least one further write operation and an increase in the estimated amount of energy required to complete the at least one further write operation. The method of claim 6 or 7 when dependent on claim 3 and / or 4, further comprising also imposing a penalty on

9. 9. The method of claim 3 or any one of claims 4 to 8 depending thereon, wherein the scheduler dynamically schedules at least a later one of the plurality of discrete time intervals in response to a read operation during or after an earlier one of the plurality of discrete time intervals.

10. 1. A holographic data storage system comprising: at least one processor configured to implement a scheduler, the scheduler configured to schedule at least one write operation over a plurality of non-contiguous write intervals, the write operation relating to a set of data to be stored in an area of ​​a holographic recording medium; an emitter system configured to emit, in each of the plurality of discontinuous writing intervals, a reference beam and an input beam in which the set of data is embedded; Holographic recording medium and arranged to expose, in each of the plurality of discontinuous write intervals, a region of the holographic recording medium to an interference pattern resulting from interference between the input beam and the reference beam, thereby performing the scheduled write operation.

11. the scheduler is configured to schedule at least one further operation in at least one interval interleaved with the plurality of non-contiguous write intervals; The system further includes one or more optical waveguide networks arranged to direct the input beam and the reference beam to the region of the holographic recording medium in each of the plurality of discrete time intervals, the system directing the reference beam using at least one of the one or more optical waveguide networks in the interleaved intervals: to another region of the holographic recording medium or to another holographic recording medium, with or without changing the angle and / or phase characteristics of the reference beam, or Different angles and / or phase characteristics, but in the same area of ​​the holographic recording medium. The system of claim 10 configured to carry

12. 12. The system of claim 11, wherein the at least one optical waveguide network has at least one configurable inductive element, and the at least one processor is coupled to the at least one inductive element to reconfigure the at least one inductive element to direct the reference beam to the other holographic recording region in the interleaved section.

13. 12. The system of claim 11, wherein the at least one optical waveguide network has at least one static guide element responsive to a controllable optical property of the reference beam, and the at least one processor is coupled to the emitter system to change the optical property of the reference beam to deliver the reference beam to the other holographic recording region in the interleaved section.

14. 14. The system of claim 11, wherein the write operation and the further write operation are associated with respective logical addresses, each logical address defining the holographic storage area and the reference beam angle and / or phase characteristics for its associated operation.

15. A scheduler embodied as executable program code stored on a computer-readable storage medium, the scheduler being configured to perform the method of any one of claims 1 to 9 when executed on at least one processor of a holographic data storage system.

Images