Waveform portability - the capacity to move a waveform from one platform to another
with minimal engineering effort - is a major advantage of software radio. However,
without employing good design practices, porting waveform software can actually
require more time than a complete redesign. Because CRT's focus is on developing
software solutions rather than complete radio products, waveform portability is
critical to our business. The following describes CRT's approach to achieving
waveform portability across platforms and some of our past experiences porting
cognitive radio software.
Platforms
Philosophically, a "platform" refers to the processing hardware resources and
operating environment of the target system. With CRT's platform independent design
approach, application specific algorithms (translated from the application problem
domain) and cognitive algorithms are designed and implemented with the goal of
platform independence to simplify the porting process.
|
System Abstraction
Achieving any degree of platform
independence requires an abstraction of
platform specific implementation details to a
more generalized model. In CRT's network or
node solutions, communication systems are
designed to follow a hierarchical structure
with each functional layer abstracted in a
logical resource perspective. This approach
permits our service and control interfaces of
the system to follow standardized
implementation approaches and interfaces
for a maximal degree of portability.
|
|
|
|
|
Cognitive Radio System Model
|
|
|
|
|
|
|
|
|
Application Portability Design
The following describes how CRT designs its software solutions as guided by the
design goals of hardware portability and simplicity and efficiency. In general, these
are competing design objectives so care must be taken to properly balance software
solutions for the envisioned applications.
Achieving Hardware Portability
Using CRT's cross-platform profile based interface, applications with specific
domain knowledge can access and claim processing resources through the
standard "brokering" mechanism. Also, using standard interpretation methods,
different hardware processing platforms can be configured and controlled to serve
the client systems that are developed independently.
Achieving Simplicity and Efficiency
In CRT's application driven system solution, the application interfaces (toward user
and toward processing platform) features simple format, open structure, and
self-explicit meta-data data interpretation. This simplifies the integration of CRT
software into existing systems and the porting of CRT software across platforms.
To minimize the inherent overhead associated with portable software, we optimize
different standard communication and interpretation methods at different functional
layers of the system. For example, standard TCP socket interconnection methods
are implemented for distributed information processing and control; and POSIX
inter-processing communication and multi-threading control are developed for
inside-node adaptation.
|
|
|
|
Example General Interface Between Application and Radio Platform
|
|
|
Example Efforts in Designing Portable Cognitive Radio Software
Case Study 1: General Radio Interface for Cognitive Radios
|
To realize a portable cognitive
radio engine, we adopted the Egg
Model architecture for interfacing
between processes for decision
based learning, gathering device
and environmental information,
and controlling radio operation.
This approach is greatly aided by
parsing XML component
descriptors (a requirement on
SCA based software radios and
increasingly common on
commercial systems) for radio
hardware, waveform information,
and environmental information.
More details about this
architecture is available in this
publication.
|
|
Case Study 2: Public Safety Cognitive Radio Node Solution
|
As described in greater detail in
this publication, we developed a
cognitive radio solution for
achieving dynamic interoperability
between public safety systems. In
this case we needed to support
two different operating
environments - Linux and
Windows - interfacing with a
common RF hardware board (a
USRP).
|
|
Case Study 3: Reconfigurable Waveform Framework for NASA STRS Radio
|
Illustrative of a more generalized
approach to SDR porting, we
adopted the control and
integrated service interface
architecture shown to the right
while supporting an SDR
development approach targeting
the AeroAstro DEVAR STRS radio
platform.
|
|
