CSU-CHILL radar's Virtual Radar System called VCHILL provides the full operation of the radar over the Internet. As a part of this project several aspects of real-time weather radar data transmission at various bandwidth levels are under investigation.

This paper presents end-system architecture for transmission of real-time digitized weather radar signals over high bandwidth network and for estimation of radar parameters. The architecture is developed based on the client-server model with the operation of multi-processes and multi-threads. The design includes the digitized radar signal acquisition, transmission, radar parameter computation, and parameter transfer, as well as generic packet and data structures. In addition, the architecture accommodates a congestion control algorithm that can be classified as the source-based rate control with User Datagram Protocol (UDP) transport protocol.

In order to provide end-users with the highest quality of data under any network conditions, a user-perspective Quality of Service (QoS) is established. Changes in the QoS of the estimated parameters with various packet loss scenarios are compared via simulation. The results show that the suitable filtering of transmitted data is effective to minimize the degradation in the quality of the computed radar parameters. The proposed data filtering policy is combined with the congestion control algorithm.

The end-system architecture for the CSU-CHILL radar was successfully implemented on both Sun/Solaris and Linux platforms. Emulated results performed over gigabit link show that the congestion control algorithm performs well and greatly improves quality of the radar display compared with the non-congestion control case.