Friday, August 24, 2018   |   Scientific Development

The Baron Processor Suite, Part 2

by Mrinal Balaji, Baron Chief Scientist

This is the second in a two-part series of articles about new hardware and technique advances in Baron radar processing. Read the first article here.

 

Paradigm-shifting Features within the Baron Processor Suite

While on the surface, the system configuration of the Baron Gen3 Processor Suite seems quite relatable to the existing processors in the radar industry, the method of execution of the various software philosophies within the processors sets them apart, raising the bar on big data processing standards to a new level. The sections that follow illustrate each of the benefits.

 

Processor Suite Internal Architecture

With the always growing amount of data, the need for frameworks to store and process this data efficiently is increasing. The Baron Processor Suite employs Data Distribution Service (DDS™), the first middleware standard that addresses challenging real-time requirements not satisfied by any other existing standard. It is capable of handling very high-performance communications and is currently employed within applications like NASA’s robotic applications, financial high speed trading, advanced telescopes, medical imaging, air traffic control, smart grid management, and several other big data applications.

DDS outlines a data-centric publisher-subscriber architecture that connects the information providers with information consumers. As opposed to traditional message oriented middleware, DDS understands and uses the context of the data to ensure that all interested subscribers have an accurate and consistent view of the data rather than simply focusing on the delivery of the message.

In addition, the architecture creates a global shared data-space that significantly streamlines integration. It transmits data directly from a publisher to all its subscribers with no intermediate servers. Publishers and subscribers can join or leave easily, be anywhere, publish at any time, and subscribe to any data they have permission to access.

DDS provides true auto-discovery of publishers and subscribers. There is no necessity to statically set IP addresses of nodes within the globally shared data-space for the purpose of message routing. It also provides several mechanisms for filtering data on sending and receiving. Subscribers/publishers determine what messages they want to receive/transmit based upon the filter they define. Timing and flow are precisely controlled while the computer operating system platform and language differences are automatically translated. This allows flexibility in using different programming languages within different processors.

 

Key Benefits of DDS 

  • Allows the processors to exchange messages internally and externally in an extremely efficient fashion.
  • Provides the ability to perform R&D on an operational system. One can develop and test an algorithm using multiple sets of setup parameters concurrently within an operational system without interfering with the operational flow of data.
  • Employs a data-centric integration model to decouple applications. Applications communicate by publishing the data they produce and subscribing to the type of data they consume. They require no knowledge of each other, only of the data they exchange.
  • Extremely straightforward to create redundant RSP (Radar Signal Processor), RCP (Radar Control Processor) or RPG (Radar Product Generator) processes without needing additional hardware.
  • Provides internal process communications through shared memory leading to extremely fast and efficient use of memory resources.
  • Provides inter-node communication through multicast and unicast. Developers can choose to use multicast or unicast based upon the publisher and subscriber relationships.
  • Server less architecture that has no single point of failure. Systems are self-healing when applications disconnect and reconnect. Automatic failover provides continuous availability when an application is no longer accessible.

 

Clutter Environment Analysis Using Adaptive Processing – CLEAN-AP™

Developed by the University of Oklahoma, Norman, Oklahoma, USA and available through an exclusive licensing agreement with Baron, the CLEAN-AP™ ground clutter filtering process allows for superior ground clutter suppression in addition to optimally and dynamically adapting the suppression process to the ground clutter environment (Warde et al. 2009, Torres et al. 2014). In development over the last several years, this new technology will be introduced within the U.S. National Weather Service WSR-88D fleet soon (Torres et al. 2012).

 

Key Benefits of CLEAN-AP 

  • CLEAN-AP™ performs automated clutter detection and suppression with no need for manual intervention.
  • The need for clutter maps is eliminated through CLEAN-AP™’s real-time ground clutter detection capability.
  • CLEAN-AP™ uses adaptive data windowing that accomplishes a good compromise between clutter suppression and data quality.
  • CLEAN-AP™ is an integrated process providing a single algorithm for ground clutter detection and filtering on a bin-by-bin basis.

The unique clutter filtering ability of CLEAN-AP™ is vital when being used as part of an integrated network. By greatly improving the accuracy of base data, benefits occur throughout the entire system, from data collection and integration to forecast models, and ultimately leading to more effective forecasts and alerting for the general public.

 

CLEAN-AP™ within the Baron Radar Processing Suite

CLEAN-AP™ processing has been implemented within the RSP and with the help of the inventors, Dr. Sebastián M. Torres and David A. Warde, the algorithm has been optimized to work at S-Band (2.7 – 3.0 GHz), High-Frequency S-Band (3.5 – 3.6 GHz) and C-Band (5.3 – 5.8 GHz) with all types of transmitters. Fielded currently at multiple operational radar sites, CLEAN-AP™ consistently proves why it is the new golden standard for ground clutter filtering in the weather industry.

Figure 1 demonstrates radar reflectivity images along H-Polarization before (Upper Panel) and after (Lower Panel) CLEAN-AP™ was applied at an operational radar site. Anomalous propagation is seen all around the radar out to a range of approximately 200 km and CLEAN-AP™ successfully and elegantly suppresses all of it in addition to the ground clutter around the radar.

Figure 2 illustrates the benefits provided by CLEAN-AP™ (Upper Panel) when compared with traditional ground clutter filtering (Lower Panel) at one of the operational sites:

  • Superior ground clutter suppression while introducing less bias in the base data.
  • Optimum amount of ground clutter removal while keeping the mixed in weather signal intact.
  • No deterioration of base data along the zero velocity isodop as opposed to noisy or missing base data with traditional ground clutter filtering schemes.
before and after CLEAN-AP application

Figure 1: Before (Upper Panel) and After (Lower Panel) CLEAN-AP™ application PPI scans. Click to enlarge.

 

Figure 2: Operational Radar with CLEAN-AP™ (Upper Panel) Vs Radar without CLEAN-AP™ (Lower Panel). Click to enlarge.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Data from the Baron Processor Suite

The versatile architecture within the various processors ensures they are able to provide base data in the following forms simultaneously:

  • Unclutter filtered
  • Ground clutter filtered but point clutter preserved
  • Point clutter filtered but ground clutter preserved
  • Ground and point clutter filtered

In addition, the data can be requested in 8-bit Integer, 16-bit Integer or 32-bit IEEE Floating Point format. Another feature available as a result of the flexible internal architecture is the ability to capture data at various points within the processing chain. The data that can be captured includes but is not limited to:

  • 32-bit IEEE floating point formatted digitized I/Q data from the IF Digitizer (IFD)
  • 32-bit IEEE floating point formatted base data from the RSP
  • 32-bit IEEE floating point formatted base data from the RCP
  • 8-bit Integer, 16-bit Integer or 32-bit IEEE floating point formatted base data from the RPG

Data can additionally be captured at various internal points within the processors (e.g., inputs and outputs of the ground clutter filtering module, inputs and outputs of the point clutter filtering module, etc.).

 

Conclusion

The field of radar processing has always been open to innovation, supported by recent advancements in computing technology, improved architectural standards, and novel feature extraction and elimination algorithms like CLEAN-APTM resulting in the culmination of the Baron Gen3 Radar processing suite which is currently operational at all Baron Gen3 operational sites.

Future work on the radar processing suite will continue to find ways to effectively blend physical models, prior knowledge, and traditional signal processing methods to address applications of increasing complexity. Novel base data capable of filling in the gaps where better understanding of physical processes is needed will also be made available to the research and operational community over time.

 

References

  • Torres S.M., Warde D.A., and Zrnić D. S., Signal design and processing techniques for WSR-88D ambiguity resolution: Part 15 The CLEAN-AP filter, National Severe Storms Lab., Norman, OK, USA, 2012.
  • Torres S.M. and Warde D. A., 2014: Ground clutter mitigation for weather radars using the autocorrelation spectral density, Atmos. Oceanic Technol., 31, 2049-2066.
  • Warde D. A. and Torres S.M., Automatic detection and removal of ground clutter contamination on weather radars, AMS 34th Conf. Radar Meteorology, Williamsburg, VA, USA, Oct. 5–9, 2009, paper P10.11.