Sang-il Kim

3D UHDTV contents production with 2/3-inch sensor cameras

Most UHDTV content is presently created using single large CMOS sensor cameras as opposed to 2/3-inch small sensor cameras, which is the standard for HD content. The consequence is a technical incompatibility that does not only affect the lenses and accessories of these cameras, but also the content creation process in 2D and 3D. While UHDTV is generally acclaimed for its superior image quality, the large sensors have introduced new constraints in the filming process. The camera sizes and lens dimensions have also introduced new obstacles for their use in 3D UHDTV production. The recent availability of UHDTV broadcast cameras with traditional 2/3-inch sensors can improve the transition towards UHDTV content creation. The following article will evaluate differences between the large-sensor UHDTV cameras and the 2/3-inch 3 CMOS solution and address 3D-specific considerations, such as possible artifacts like chromatic aberration and diffraction, which can occur when mixing HD and UHD equipment. The article will further present a workflow with solutions for shooting 3D UHDTV content on the basis of the Grass Valley LDX4K compact camera, which is the first available UHDTV camera with 2/3-inch UHDTV broadcast technology.

A location inference algorithm based-on smart phone user data modelling

Human life is rapidly changing around the world in terms of smart phones as the number of users is increasing in a surprising rate and the recent development of remarkable performances. And, there is wide range of smart phone services available based on the location information. However, levels of location based services currently provided are not satisfactory due to the GPS error rate. Though there have been many studies offering measurement technologies based on a variety of network infrastructure or technologies to compensate GPS errors, there still exist disadvantages such as large error rates indoors, energy consumption and need of additional infrastructure. Nowadays, location inference study using available user log data on smart phones, not the measurement technology based on a network, is being made to improve these shortcomings. Smart phone has a variety of user log data. Also, it can obtain variety of information about the surrounding environment around the user using the embedded hardware sensors. These user log and collected sensor data can be modelled as an ontology model then meanings of the information can be apprehended to infer the user location. This paper presents an algorithm to infer user location through an ontology modelling of the collected sensor data and the logged user information such as time, day of the week, location, and kind of applications used by user.

A BGP session takeover method for high availability based on virtualization technique

Recently, the improvement in the data communication networks in terms of availability and reliability is emerging as a critical issue. In this context, the high availability and reliability of routers that control data flow and routing path between networks are being recognized as critical problems. Many methods for providing high availability to minimize the loss cost caused by the failure of a router are being studied. This paper analyzed the takeover method proposed in the previous studies to support the high availability of the BGP protocol, and proposed a BGP session takeover method using the virtualization technique to improve the takeover performance.