10-01-2012, 11:36 AM
Co-Design of Digital Telecommunication System
[attachment=15936]
Why codesign of Telecom System?
The rapid breakthrough of consumer electronics (CD, DCC, DAB, ….), wireless or wired voice and data networking (ISDN, GSM, Videophone, …), broadband networks (ATM, ADSL, …) and multimedia is phenomenal these days.
The digital communication technique is the basis of all these fast growing industrial activities.
Development of digital communication is possible due to combined growth of VLSI and DSP.
DSP System: Performs real-time mathematical transformation on digitized samples of analog signal with finite bandwidth and signal to noise ratio (SNR).
These transformation can be implemented either on a programmable processor using software, or application-specific hardware and determined by trade-offs between cost, power, performance and flexibility.
Codesign of Telecom system
DSP based products have a growth rate of more than 35% per year.
The average time to market window is reduced to few months only.
Complexity and functional density is on demand.
Design productivity:
communication system designer conceive the design at board and executable concurrent programmable paradigm that is not understood by chip architect who works in the RTL domain. Gap between system design and implementation.
The system need to be design at processor-memory level by reusing component designs. This needs methodology and codesign approach.
Size of design team does not seem to grow as chip complexity grows.
Hence, there is a need to increase design productivity and seamless transition of design strategy from software centric to implementation-reuse hardware-software codesign approach.
Specification view of DSP system
Digital Signal results from: binary encoding of time and range discretized measurable continuous time, continuous range quantities.
Sampling occurs at or above Nyquist frequency and coding is done with just enough word-length to maintain SNR.
Digital signals are usually fixed-point type to save power and hardware to meet desired performance.
Digital signals are stream of digital words due to periodic sampling. These words are naturally structured into multidimensional arrays, which are to be processed frame period Tf, that is the duration of the algorithm when it takes a set of input to produce the set of results.
Thus the elementary DSP algorithm is a dataflow function
[attachment=15936]
Why codesign of Telecom System?
The rapid breakthrough of consumer electronics (CD, DCC, DAB, ….), wireless or wired voice and data networking (ISDN, GSM, Videophone, …), broadband networks (ATM, ADSL, …) and multimedia is phenomenal these days.
The digital communication technique is the basis of all these fast growing industrial activities.
Development of digital communication is possible due to combined growth of VLSI and DSP.
DSP System: Performs real-time mathematical transformation on digitized samples of analog signal with finite bandwidth and signal to noise ratio (SNR).
These transformation can be implemented either on a programmable processor using software, or application-specific hardware and determined by trade-offs between cost, power, performance and flexibility.
Codesign of Telecom system
DSP based products have a growth rate of more than 35% per year.
The average time to market window is reduced to few months only.
Complexity and functional density is on demand.
Design productivity:
communication system designer conceive the design at board and executable concurrent programmable paradigm that is not understood by chip architect who works in the RTL domain. Gap between system design and implementation.
The system need to be design at processor-memory level by reusing component designs. This needs methodology and codesign approach.
Size of design team does not seem to grow as chip complexity grows.
Hence, there is a need to increase design productivity and seamless transition of design strategy from software centric to implementation-reuse hardware-software codesign approach.
Specification view of DSP system
Digital Signal results from: binary encoding of time and range discretized measurable continuous time, continuous range quantities.
Sampling occurs at or above Nyquist frequency and coding is done with just enough word-length to maintain SNR.
Digital signals are usually fixed-point type to save power and hardware to meet desired performance.
Digital signals are stream of digital words due to periodic sampling. These words are naturally structured into multidimensional arrays, which are to be processed frame period Tf, that is the duration of the algorithm when it takes a set of input to produce the set of results.
Thus the elementary DSP algorithm is a dataflow function