`
`DECLARATION OF YONATAN LAVI
`
`1. My name is Yonatan Lavi. I live in Tel Aviv, Israel. I am providing this
`
`declaration based on my personal knowledge and I am not being
`
`compensated for providing this declaration.
`
`2. I have been informed by attorneys for Microsoft Corporation (“Microsoft”)
`
`that Bradium has asserted several patents which list me as a co-inventor,
`
`along with Isaac Levanon. I understand that these patents include U.S.
`
`Patent Nos. 7,139,794 B2 (“the ‘794 Patent”), 7,908,343 B2 (“the ‘343
`
`Patent”), 8,924,506 B2 (“the ‘506 Patent), and 9,253,239 B2 (“the ‘239
`
`Patent”). I understand that all of these patents are related to six provisional
`
`U.S. patent applications filed on December 27, 2000, which also list myself
`
`and Mr. Levanon as co-inventors. In this declaration, I will discuss my
`
`knowledge of 3DVU, Inc. and 3DVU, Ltd. (collectively, “3DVU”),1 which
`
`originally filed these patent applications.
`
`3. I was first hired to work at 3DVU (then known as GACentral.com, Inc.) by
`
`Isaac Levanon in mid-1999 as a software developer. I, along with two other
`
`
`1 3DVU was originally formed as GACentral.com, Inc. and changed its name to
`FlyOver Technologies, Inc. in early 2000 before changing its name again to
`3DVU. 3DVU, Ltd. was set up as a subsidiary of 3DVU, Inc., which was a US
`company. To avoid confusion, I will refer to all of these companies as “3DVU”
`unless the difference is relevant to a particular issue.
`
` 1
`
`
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`developers, Ohad Eder-Pressman and Eyal Navon, started working
`
`informally for Mr. Levanon and his company (3DVU) shortly after I
`
`graduated from high school. The three of us worked essentially as
`
`independent contractors (although there was no formal employment or other
`
`contract at the time) for a few months spanning the years 1999-2000. My
`
`recollection is that Mr. Levanon communicated mostly with Mr. Eder
`
`Pressman, although the three of us were paid individually. All three of us
`
`were offered (and I accepted) shares of stock in the company in the year
`
`2000. Mr. Eder-Pressman and Mr. Navon were also recent high school
`
`graduates with some software development experience who were contracted
`
`by Mr. Levanon to write code for 3DVU. Mr. Levanon’s role was primarily
`
`running the company, rather than in developing software. Mr. Levanon was
`
`not a computer programmer, and my recollection is that at that time and
`
`throughout the time that I worked at 3DVU he did not have the ability to
`
`write software code himself.
`
`Original 3DVU prototype and 2000-2001 patent applications
`
`4. When I initially worked for 3DVU in 1999 and 2000, 3DVU was primarily
`
`focused on developing a prototype of a website for visualizing aerial
`
`imagery for an area such as an airport. In particular, we wanted to develop a
`
` 2
`
`
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`website that might be used to show pilots a perspective view of the area
`
`around an airport, which was the scenario that many of our early efforts
`
`were designed around. The “GA” in the original name of the company was
`
`short for “general aviation.” Mr. Levanon wanted the prototype to offer
`
`pilots the ability to simulate flying over an aerial photograph of an airfield.
`
`For example, Ex. B to a declaration that Mr. Levanon filed on January 10,
`
`2006 during the prosecution of Application No. 10/035,981 (which became
`
`the ‘794 Patent) shows how the “FlyOver” imagery was part of a website
`
`showing information about an airfield:
`
`
`I have attached a true and correct copy of the same document to this declaration as
`
`
`
` 3
`
`Exhibit A.
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`5. The “FlyOver” imagery in the 1999-2000 prototype basically took a two-
`
`dimensional image (like an aerial or satellite photograph) and transformed it
`
`(performing trigonometric calculations to determine where pixels from the
`
`image would appear on the screen) to create a view from a simulated
`
`perspective. The process of transforming a 2D image to view it from a
`
`three-dimensional perspective was already very well-known at the time,
`
`such as in texturing in video game applications. This original prototype also
`
`did not include any elevation data or way to visualize elevation data, so any
`
`scene that a user looked at with the “FlyOver” viewer would have looked
`
`like a flat plain or plateau, even if the user was looking at satellite or aerial
`
`photographs of a place with hills and valleys or other terrain.
`
`6. The prototype also used a technique called “MIP-mapping,” which is a basic
`
`computer graphics concept that I knew was well-known long in the industry
`
`before any work that we did at 3DVU. For example, the OpenGL software
`
`rendering library, which was developed by Silicon Graphics, used MIP-
`
`mapping extensively. In MIP-mapping, a source image is first divided up
`
`into a series of sub-images, which we sometimes referred to as image parcels
`
`or tiles. These sub-images are then down-sampled into a series of derivative
`
`images at progressively lower resolution. Progressively improving the
`
` 4
`
`
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`resolution of an image by gradually replacing lower-resolution mip-mapped
`
`versions of the image with higher-resolution versions of the image was also
`
`well-known. For example, many video games used mip-mapped textures to
`
`whose resolution would be progressively enhanced as the viewpoint
`
`approached a portion of the scenery. In fact, as I discuss further, the original
`
`3DVU prototype used a mip-mapped file format called FXT1 which was
`
`originally developed by another company for texture compression in video
`
`games.
`
`7. Although the overall process was not new, we did develop what we thought
`
`were some improvements to the way that image parcels in a MIP-map could
`
`be stored on a server and requested by a client. The 3DVU prototype in
`
`1999-2000 stored the image data for a large set of tiles within a single large
`
`file (a master index file) that had a specified format. The idea of storing
`
`smaller bits of data within a larger file was also something that was well-
`
`known before we built this prototype, but most systems used a “lookup
`
`table” at the beginning of the file that listed the “offset” (distance from the
`
`start of the file, measured in bits or bytes) to the desired data. Ohad Eder-
`
`Pressman, who worked with us at 3DVU, figured that the system could skip
`
`using the lookup table if all of the tiles within the file were the same byte
`
` 5
`
`
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`size. If the tiles were all the same size, the client device could simply
`
`calculate the offset based on the X, Y position of the tile and the level of the
`
`tile within the hierarchy of resolutions. Calculating offsets in this manner
`
`also required a variable in the code (TileSize) for the size of the tile.
`
`8. We described this method of calculating offsets for fixed byte size tiles in a
`
`Declaration that Mr. Levanon and I submitted to the U.S. Patent and
`
`Trademark Office on June 13, 2006 in support of the application that became
`
`the ‘794 Patent. As we explained on page 3/110 of that Declaration:
`
`9. In the same declaration, we also provided an example of the code which the
`
`prototype used. The code at page 20/110 of this declaration shows how to
`
`calculate offsets within the master index file to locate particular tiles:
`
`
`
`
`
` 6
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`
`10. This source code used a formula to calculate the byte offset for a particular
`
`tile based on the size of each tile and the X, Y position and level of each tile.
`
`11. As noted above, we also used compression to reduce the size of each data
`
`block corresponding to an image parcel (tile). However, in order to preserve
`
`the system of locating tiles by calculating offsets rather than using a lookup
`
`table, we used a variation of a texture compression format, known as FXT1,
`
`which had a fixed compression rate per pixel, so that a tile of a given size
`
`measured in pixels would always be the same size as measured in bytes.
`
`
`
` 7
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`FXT1 was an open source texture compression format released by a
`
`company called 3dfx in around 1998-1999, designed mainly for video
`
`games. We (at 3DVU) did not invent this file format, we simply used it
`
`(with slight modifications) for our application.
`
`12. For example, as we mentioned on page 7/110 of the June 13, 2006
`
`Declaration:
`
`13. The requirement for fixed byte size tiles based on fixed ratio compression is
`
`
`
`also reflected in the patents that were eventually filed based on the
`
`provisional applications that 3DVU filed in December 2000. The ‘343
`
`Patent states at column 6, lines 19-22 that “[t]he preferred compression
`
`algorithm may implement for example a fixed 4:1 compression algorithm
`
`such that each compressed and stored image parcel has a fixed 2K byte
`
`size.” The same sentence appears in the ‘506 Patent at column 6, lines 25-
`
`28 and in the ‘239 Patent at column 6, lines 21-24. All of the independent
`
`clams of the ‘343 Patent and ‘506 Patent also say that the tiles have to be
`
`
`
` 8
`
`fixed byte size.
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`14. One of the intended purposes of the prototype that we built in 1999-2000
`
`was to operate over “narrowband” communication channels, or
`
`communication channels with limited bandwidth. For example, the
`
`prototype that we built in 1999-2000 used a dial-up modem connection, and
`
`such modems at the time typically offered speeds such as 56 KBps, 28.8
`
`KBps, 14.4 KBps, or even lower.
`
`15. In our design of the prototype, limits on processing power and the limits on
`
`bandwidth were distinct issues. For example, the size of the tiles was an
`
`aspect of the design that affected the bandwidth over which the system could
`
`successfully operate, while the manner in which tiles were rendered and
`
`displayed was a design consideration that could affect how much computing
`
`power was required to successfully operate the device. While these
`
`considerations could occur in the same device, they were also independent
`
`of each other (for example, if a computer with a powerful processor was
`
`connected to the Internet with a very slow connection, or if a device with
`
`limited processing and rendering power was connected to the Internet with a
`
`fast connection) and I did not consider limited bandwidth and the processing
`
`power of the device (or the size or shape of the computing device) to be
`
` 9
`
`
`
`synonymous or interchangeable.
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`16. We (at 3DVU) did not attempt to operate any of the software on a Personal
`
`Digital Assistant (PDA) or other mobile computer system, or otherwise write
`
`special adaptations in the code specific to mobile operating systems, before I
`
`left for the IDF, or before the first provisional applications for patents
`
`relating to this prototype were filed in December 2000. My recollection is
`
`that the testing of our prototype in 1999-2000 was all on a browser interface
`
`operating on a standard home desktop personal computer. I do not recall
`
`exactly what model of computer or computers we used, but to the best of my
`
`recollection, the computers that we used for our prototype would have most
`
`likely had a processor such as an Intel Pentium II or III or a processor of
`
`similar computing power.
`
`17. The software used in the original prototype, both for the server and for the
`
`client, was written primarily in standard C++ code. The original software
`
`was not written specifically for a mobile or embedded system. However,
`
`adapting the software for a PDA later on would have been a straightforward
`
`matter because the Windows CE (later Windows Mobile) mobile operating
`
`platform was designed to maximize the overlap in functionality with the
`
`Windows desktop environment, precisely so that developers could easily
`
`adapt Windows desktop applications for mobile use. The original software
`
`
`
`
`10
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`also used standard, pre-existing network protocols to send and receive data,
`
`and the code was not specialized for the network layer.
`
`18. I do not believe that we ever sold or licensed the “flyover” prototype.
`
`Additionally, significant development work and many changes to the code
`
`and the software took place between the original “flyover” prototype and the
`
`products that 3DVU worked on in later years, such as the Denso car
`
`navigation system and the Navi2Go app, which I will discuss further in this
`
`declaration.
`
`19. I left 3DVU in March 2000 in order to begin my compulsory service in the
`
`Israel Defense Forces (IDF). I served in the IDF as a scientific programmer
`
`with the unit 8200, which is the Israeli Intelligence Corps unit responsible
`
`for collecting signal intelligence and code decryption. I served in the IDF
`
`for approximately 3 years. During the time that I was in the IDF, I had
`
`occasional contact with Isaac Levanon either when required me to sign
`
`something or when someone at 3DVU had a technical question. However,
`
`since I was on active duty in the IDF, my focus was on my military duties.
`
`3DVU’s role in vehicle navigation products
`
`20. I returned to 3DVU full-time when my military service was complete in
`
`March 2003. At this time, 3DVU was working with a company called
`
`
`
`
`11
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`Denso, which had a line of car navigation products. These Denso car
`
`navigation products used satellite or aerial imagery to show the route, in
`
`addition to simple maps. 3DVU’s role in the development of this product
`
`consisted of processing the image data that was used in the navigation
`
`product, and rendering it with 3d perspective into a bitmap layer in real time.
`
`The user interface was designed primarily by Denso and would interact with
`
`the 3DVU portion of the software by making calls to an Application
`
`Programmer Interface (API), which 3DVU provided. Denso sold the final
`
`products in Japan under either the Kenwood or Daewoo brand names.
`
`21. The final car navigation products consisted of a self-contained in-car
`
`navigation system, which stored all navigation data on a hard drive. This
`
`system did not download data over the Internet, or any other network. There
`
`were two major generations of these car navigation products released in
`
`2002 and 2003. These car navigation products did not include any digital
`
`elevation model or digital terrain model. In other words, these products only
`
`showed the user a perspective view of flat imagery, without any
`
`representation of terrain. For example, a video on Mr. Levanon’s YouTube
`
`account (available at https://www.youtube.com/watch?v=3btUNuCYM6M)
`
`shows what the early version of the Denso system looked like in operation,
`
`
`
`
`12
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`and it is clear out several points in the video (note the horizon at, for
`
`example, 1:27 and 2:06-2:25) that the imagery being displayed only shows
`
`the imagery as a flat surface, and does not represent the terrain with an
`
`elevation model:
`
`
`
`
`22. One significant difference between the first and second generations of the
`
`car navigation products is that the first generation used fixed ratio
`
`compression, while the second generation used variable ratio compression
`
`based on JPEG. JPEG compression effectively stores a series of numbers
`
`
`
`
`13
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`representing the mathematical characteristics of waves that can be used to
`
`reconstruct an image, rather than a pixel-by-pixel bit map of an image.
`
`Compressing different images using JPEG, even if the uncompressed images
`
`are the same size in pixels and bit depth per pixel, will result in compressed
`
`images that have different byte sizes. This is referred to as “variable
`
`compression ratio.” Variable ratio compression allowed us to use much less
`
`memory than fixed ratio compression because variable compression achieve
`
`higher average compression ratios. This was important in the car navigation
`
`products for Kenwood because the entire navigation database was stored on
`
`a hard drive, and therefore efficient use of memory was a significant
`
`consideration. After we began using variable ratio compression in 3DVU
`
`products, we did not go back to using fixed ratio compression for any later
`
`3DVU product.
`
`Improvements to 3DVU technology after 2000-2001
`
`23. After I returned to 3DVU from the IDF, I started working on an improved
`
`algorithm for digital elevation modeling. I do not recall any development of
`
`digital elevation modeling before I left for my military service in 2000, and I
`
`am unaware if 3DVU had started experimenting on digital elevation
`
`modeling during the time that I was gone for my military service. However,
`
`
`
`
`14
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`no digital elevation modeling had been incorporated into the car navigation
`
`product before I finished my military service and came back to 3DVU as an
`
`employee. I spent a large amount of my time after returning from my
`
`military service working on improving the digital elevation model.
`
`24. The digital elevation model basically consisted of a database containing
`
`elevation data for geographic regions. A digital terrain model is a 3-D
`
`representation of terrain that uses the coordinates of points in space (based
`
`on some sort of mapping data containing elevations) as the vertices for a
`
`model that represents the surface of the earth. A well-known challenge in 3-
`
`D modeling, including modeling terrain surfaces, was how to manage digital
`
`terrain models at varying levels of detail. The problem is roughly analogous
`
`to the level of detail challenge in displaying textures. For example, a digital
`
`terrain model at a high resolution (that is, with a high number of vertices)
`
`can take a lot more system resources to load or download. However, a
`
`digital terrain model at a low resolution (that is, with a low number of
`
`vertices) will look blocky and unnatural viewed close up. There were a
`
`number of methods already known for building 3-D meshes at various levels
`
`of detail based on viewpoints, for example, by adding additional vertices as
`
`the viewpoint gets closer to the mesh.
`
`
`15
`
`
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`25. I worked on developing a specific algorithm for building 3-D meshes at
`
`various levels of detail that was intended to be less computationally
`
`intensive than previous methods. I believed that I was successful in
`
`developing a new algorithm for displaying 3-D meshes, such as terrain, and I
`
`described this algorithm in a patent application filed on February 8, 2006,
`
`which eventually issued as US patent number 7,561,156 B2 (“the ‘156
`
`patent”) on July 14, 2009. The work that I invested in developing this
`
`algorithm, reflected in the ‘156 patent, all took place after I returned from
`
`the IDF in 2003, and not before I left for the IDF in 2000. Consequently, the
`
`digital terrain modeling algorithm that I developed is not reflected in the
`
`patents originally filed in 2000 and 2001, because I had not developed it yet.
`
`26. We started introducing my new 3-D digital terrain model into 3DVU
`
`technology toward the end of 2005. For example, a 3DVU press release
`
`dated November 22, 2005, which I accessed via the Internet Archive at
`
`https://web.archive.org/web/20060210012618/http://www.3dvu.com/pdf/Da
`
`ewoo__pr.pdf, announced that the third generation of the 3DVU mapping
`
`platform would include “full landscape elevation.” The same press release
`
`includes a quote from Mr. Levanon stating that “for the first time, navigating
`
`users will view realistic imagery with the elevation of mountains and valleys
`
`
`
`
`16
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`on their mobile devices.” This press release and statement are consistent
`
`with my recollection that the 3-D terrain model was introduced around 2005.
`
`A true and correct copy of this press release, as shown on the Internet
`
`Archive, is attached as Exhibit B.
`
`Discussions with Microsoft
`
`27. I was aware from discussions with Isaac Levanon that Mr. Levanon had
`
`some discussions with Microsoft in 2005 because Mr. Levanon wanted to
`
`sell the company. However, Mr. Levanon did not provide a significant
`
`amount of detail to me about those negotiations. I recall Mr. Levanon
`
`asking me some very high-level technical questions during this time period,
`
`which I understood to have been in response to inquiries from Microsoft.
`
`These questions did not go into any significant technical detail about the
`
`operation of our technology, and only provided a very rough description of
`
`what our software was ideally capable of.
`
`28. At the time that these discussions took place, 3DVU only had two or three
`
`developers working for it, including myself. Another developer named Lion
`
`Burger came to 3DVU during the period when I was serving in the IDF and
`
`left in 2007. 3DVU hired another developer around this time, although I do
`
`not recall if this individual began work before or after the discussions took
`
`
`
`
`17
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`place, and this developer only worked at 3DVU for about a year and a half.
`
`3DVU’s headquarters was listed as Mr. Levanon’s home in in Ra’anana,
`
`Israel, while the actual office that we worked from in working hours was an
`
`apartment in a residential building in Herzelia, Israel. I recall that Mr.
`
`Levanon’s wife and children may have had some formal title within the
`
`company, but I did not see them regularly at the office, with the exception of
`
`Mr. Levanon’s daughter Mor Levanon, who began showing up
`
`approximately around 2007.
`
`29. While Mr. Levanon was in discussions with Microsoft, Mr. Levanon
`
`occasionally sent me questions about 3DVU’s development process which
`
`may have originated from Microsoft, although Mr. Levanon generally sent
`
`me these questions by email in a way that did not indicate where the
`
`question came from. These questions were somewhat awkward and difficult
`
`to answer, because 3DVU at the time did not really have an organized
`
`software development process due to the fact that there were only two or
`
`three developers working there at a time. At one point during this period, I
`
`got on a phone call with someone from Microsoft at Mr. Levanon’s request.
`
`In the phone call I was asked to give a technical overview about the state of
`
`software development in the company and the existing capabilities of our
`
`
`
`
`18
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`software library, although not any technical details of the operation of that
`
`technology. My ability to answer questions during this conversation was
`
`extremely limited because I had almost no experience at the time in spoken
`
`English. I do not believe that either the information that I provided in the
`
`questions that I answered for Mr. Levanon, or the information that I
`
`provided in the phone call with Microsoft, would have been at a sufficient
`
`level of technical detail to provide any useful information about how to re-
`
`create our technology at that point. I am also unaware of any source code or
`
`other similar detailed technical information being provided to Microsoft
`
`during this period.
`
`30. At the time of the negotiations with Microsoft, there were issues with the
`
`3DVU software that would have presented significant challenges if we had
`
`attempted to scale them. For example, our software at the time required
`
`preprocessing of image, terrain, and other geographic data, which we
`
`typically retrieved from commercial sources, before it was ready to load
`
`onto a server. However, our preprocessing software was not yet
`
`parallelized, i.e., set up so that processing could be performed
`
`simultaneously by different processing elements. With large scale maps, the
`
`issue was processing the data in such a way that would avoid conflicts or
`
`
`
`
`19
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`errors along boundaries between adjacent data processed by different
`
`processing elements. We needed to solve this problem in order for the
`
`software to be realistically scaled up to display a worldwide map. This
`
`particular problem took me working alone much of the next two years, into
`
`2007 or 2008. Additionally, the application programmer interface (API) (the
`
`set of definitions, protocols, and tools used for working with our software)
`
`was not very mature or user-friendly, which meant that it would have been
`
`difficult for a programmer from an outside company that acquired us to
`
`write applications to work with our mapping engine and expect them to
`
`work.
`
`Navi2Go
`
`31. 3DVU also worked on a product called Navi2Go in the later years that I
`
`worked at 3DVU. While some of my work on image preprocessing and the
`
`visualization library related to this product, the Navi2Go product was mostly
`
`designed by other developers at 3DVU. Navi2Go was generally designed to
`
`allow users to look up directions and retrieve satellite imagery covering the
`
`area of a route using a web application, then load the route instructions and
`
`imagery onto a PDA or other mobile device. Once the route and imagery
`
`were downloaded, either the web app or the mobile device/PDA could
`
`
`
`
`20
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`display the imagery along the route from a perspective view, much like the
`
`car navigation products. The primary way that Navi2Go was designed to be
`
`used was in this manner, in which the user first downloaded the routing
`
`information and imagery onto a desktop computer, then transferred this
`
`information on to a portable device (for example, using a memory card of
`
`some type or a USB drive).
`
`32. Navi2Go was also primarily written in standard C++ code language. There
`
`was a separate project involving code which was written for Java2ME,
`
`which is an operating environment based on the Java programming language
`
`and was commonly used for small cell phones (also known as “feature
`
`phones”) at the time. However, this project did not include any digital
`
`elevation or terrain modeling features, and would have only displayed
`
`imagery of a region as a “flat” projection, in a similar manner to the 1999-
`
`2000 prototype. Although the Java-based product and the C++ code
`
`libraries generally did not (and could not) share software infrastructure,
`
`some of the operation of this software shared similarities with the C++ code
`
`at a high level, such as the image decompression and the high level structure
`
`
`
`
`21
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`of the rendering software. For example, the image below,2 which to the best
`
`of my recollection is a fair and accurate depiction of the appearance of the
`
`web application user interface at the time we introduced the Navi2Go
`
`application, includes a “save-to-PDA” option in the upper right hand corner
`
`of the screen:
`
`
`33. We developed an algorithm in order to allow the web application to most
`
`efficiently retrieve in advance the imagery that would be needed for a
`
`particular route. After the user’s route was calculated, the algorithm
`
`determined which tiles of the underlying imagery database were intersected
`
`by the routes, and calculated which tiles to download so that higher
`
`
`2 This image was retrieved from a news article about Navi2Go at
`http://bgr.com/2007/12/08/3d-roadmapping-with-navi2go/, which I have also
`attached as Exhibit C.
`
`
`
`
`22
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`resolution imagery would be download for areas closest to the routes and
`
`lower resolution imagery would be downloaded to show areas that would
`
`appear more distant in the background, but would not to be shown close to
`
`the viewing perspective. The amount of high-resolution imagery download
`
`by the software could vary based on the available memory on the portable
`
`device which it would be downloaded to.
`
`34. We described this algorithm in a provisional U.S. patent application (No.
`
`60/880,674) filed in January 2007, and a non-provisional U.S. patent
`
`application (No. 12/015,068) which was published in 2008. The published
`
`application (US 2008/0294332 A1) is attached as Exhibit D. For example,
`
`paragraph 9 of the published application says that “the purpose of this
`
`invention is to provide a solution for the off-line visualization of memory-
`
`consuming datasets such as aerial photography, terrain elevation and 3D
`
`buildings, within the storage and connectivity limitations of today’s
`
`mobile/portable computing platforms as navigation products, such as
`
`Personal Navigation Devices (PND’s).” Additionally, Fig. 4c shows an
`
`example of how the application would download higher-resolution tiles
`
`closest to a route and lower-resolution tiles further away:
`
`
`
`
`23
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`
`35. Some versions of Navi2Go also allowed a user to download the same
`
`information onto a mobile device over a cell phone network in real time;
`
`however, there were some significant problems with this operating mode
`
`which made it generally impractical for most users. First of all, the speed of
`
`wireless networks at the time made the real-time downloading feature
`
`generally very sluggish, choppy, and unsatisfying. Second, the large amount
`
`of data that needed to be downloaded, including imagery, meant that users
`
`who used the real-time downloading feature would incur very large data
`
`usage charges from their cellular carrier. Downloading the data first onto a
`
`
`
`
`24
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`desktop computer, then transferring it to a portable device avoided both of
`
`these problems.
`
`36. Like every other implementation of 3DVU’s technology by 2007, Navi2Go
`
`used variable ratio compression (specifically JPEG) to transmit and store
`
`imagery. As I previously explained, variable compression ratios allowed
`
`more efficient compression than the fixed compression ratio, fixed byte size
`
`tile techniques that 3DVU used in 1999-2000 before I left for the IDF.
`
`Therefore, reverting to the old fixed compression ratio techniques would
`
`have meant that the imagery would require more storage or more bandwidth
`
`than with variable compression, which would have been significantly
`
`disadvantageous either for storing downloaded data on a mobile device or
`
`downloading it in real time.
`
`37. Although I am aware that Navi2Go received some attention from trade
`
`shows and trade papers, my impression was that this program did not
`
`experience significant success. For example, 3DVU never grew from its
`
`headcount of 2-3 developers at any given time from the time that Navi2Go
`
`was released until I left 3DVU, which I would have expected if the software
`
`had done very well commercially.
`
`
`
`
`25
`
`Microsoft, Ex. 1017
`Microsoft v. Bradium, IPR2016-00449
`
`
`
`Compensation for working at 3DVU
`
`38. Toward the end of my time at 3DVU in 2008-2009, I began to notice
`
`irregularities in my pay statements. I noticed that my pay suddenly became
`
`adjusted (without my permission) so that most of my compensation was
`
`characterized as a bonus rather than a salary, which reduced the amount that
`
`3DVU would have needed to pay in various government-required salary
`
`withholdings, which were based on salary only. Additionally, I began to
`
`experience delays receiving my paychecks. My impression at the time was
`
`that 3DVU was struggling to maintain its status as a financially viable
`
`business. I left 3DVU in August 2009 because, in addition to 3DVU’s
`
`evident financial difficulty, it had become apparent to me that the company
`
`was not growing and it did not appear to me that it was likely to grow or
`
`have increased commercial success in the future. I also never received my
`
`final paycheck for working at 3DVU after I told Mr. Levanon tha