Optimal Determination of Common for SDR C. Moy, J. Operators for - - PowerPoint PPT Presentation

▶
optimal determination of common
SMART_READER_LITE
LIVE PREVIEW

Optimal Determination of Common for SDR C. Moy, J. Operators for - - PowerPoint PPT Presentation

Oct 04, 2023 38 likes •200 views

Architecture optimisation Optimal Determination of Common for SDR C. Moy, J. Operators for Multi-Standard Software Palicot, Virgilio RODRIGUEZ, D. Giri Defined Radio Overview 4th Karlsruhe Workshop on Software Radios Mathematical

slide-1
SLIDE 1

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Optimal Determination of Common Operators for Multi-Standard Software Defined Radio

4th Karlsruhe Workshop on Software Radios Karlsruhe, Germany

  • C. Moy
  • J. Palicot

Virgilio RODRIGUEZ

  • D. Giri

SCEE Group, IETR/Supélec Cesson-Sevigné, France email: vr@ieee.org

23 March, 2006

slide-2
SLIDE 2

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Outline

1

Overview

2

Mathematical framework

3

Architecture optimisation

4

Discussion / Outlook

slide-3
SLIDE 3

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Problem statement

The optimal design of a multistandard reconfigurable radio is the right choice between two extremes:

One extreme: the “Velcro” solution (one self-contained complex module for each supported standard) Other extreme: install only the most “primitive” components (adders, multipliers, etc), and provide ”higher level” functionality through multiple calls

Trade-off:

Velcro architecture generally provides best performance, but at highest manufacturing cost (and size/weight) Other extreme likely minimises cost (& size/weight) but at unacceptable performance (multiple calls add latency!)

Our approach : build a mathematical framework to find the optimum between these extremes

slide-4
SLIDE 4

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Overview

We model the reconfigurable radio as a (hyper)graph of progressively simpler functional modules The functionality of a given module can be provided in 2 ways:

installing a dedicated component optimised for that task invoking (repetitively) lower level modules

With each module we associate 2 “costs”: monetary and computational (delay) When a lower-level module is needed several times it is invoked multiple times (not physically replicated ) The cost of a design is a weighted sum of the totals of both costs To find the optimum, we use: (1) exhaustive search & (2) simulated annealing

slide-5
SLIDE 5

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

A graph for a tri-standard radio

slide-6
SLIDE 6

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

A realistic “sub-design” example

Want an architecture to support 3 main functional modules: OFDM, Equalisation, and Multichannel processing Presumably these modules are part of grander design Equalisation (to compensate for multipath) can be implemented via

FIR filtering FFT (great for channels with long impulse responses)

Multichannel refers to channelisation function of BS (needs to process many channels in parallel). Two

  • ptions:

“Classical” channel per channel Filter bank channeliser (which can be implemented via FFT)

slide-7
SLIDE 7

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Sub-graph of design choices I

slide-8
SLIDE 8

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Performing the optimisation

Key question: should we install a self-contained/dedicated component to perform a given functionality, or should we invoke lower level modules/components? Each component is characterised by 2 “costs”: monetary, and “computational” (time) When a lower-level module is needed several times it is invoked multiple times (not physically replicated ) Choose components to minimise a weighted sum of total monetary plus total computational costs Algorithms:

Exhaustive search (“brute force”) Simulated annealing

slide-9
SLIDE 9

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Sub-graph with some parameters

slide-10
SLIDE 10

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Results

Results are heavily influenced by chosen weights (monetary vs. computational)

when “delay” costs weigh heavily, complex, expensive but high-performing dedicated components are chosen when “delay” costs weigh less, simpler, reusable components are chosen (leading to a less expensive design but with higher latency)

Above agrees with intuition

slide-11
SLIDE 11

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

An optimal design

slide-12
SLIDE 12

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Discussion

We presented a mathematical framework to find an

  • ptimal architecture for a multistandard reconfigurable

radio Key: graph of progressively simpler functional modules, showing their interdependencies (AND, OR ) Key question: install (specialised component) or invoke (lower levels)? Choose components to minimise weighted sum of 2 “costs”: money and delay A realistic “sub-design” has been solved both by “brute force” and by simulated annealing Results are highly influenced by weights, and are intuitive

slide-13
SLIDE 13

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

Immediate Future (in progress)

Re-building the hypergraph of design choices. Researchers seek:

new operators that may be common to several communication blocks to replace time-domain with new frequency-domain algorithms (which would add arcs pointing to FFT ) to include more communication standards in the graph, and track their evolution

Change objective function to minimise (monetary) cost

  • nly, subject to delay constraints (“deadlines”)

Transform the architecture optimisation into a “network design problem” (to access extensive literature with many algorithms and results)

slide-14
SLIDE 14

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

In the more distant horizon

Consideration of: multiple instances of same component (butterfly, FFT, etc) to reflect real market choices time needed to re-configure the radio while switching standards “travel time” of signals from a component to another possible contention among high level modules for the service of the same lower-level module (which may be critical if the SDR needs to support simultaneous

  • peration over several standards)
slide-15
SLIDE 15

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

A glance into the future: graph/network

slide-16
SLIDE 16

Architecture

  • ptimisation

for SDR

  • C. Moy, J.

Palicot, Virgilio RODRIGUEZ,

  • D. Giri

Overview Mathematical framework Architecture

  • ptimisation

Discussion / Outlook

A glance into the future: graph/network

FFT MultiChan Equalis OFDM FiltBk ChPCh ButterF 0/s €/0 €/0 0/s €/0 €/0 0/s FIR CIC Cordic Map 0/s 0/s €/0 €/0 €/0 0/s 0/s 0/s 0/s

  • rigin