September 23-24, 2010 San Jose, California, USA Baseband Fading Channel Simulator For Inter-Vehicle Communication Using SystemC-AMS Abdelbasset Massouri Antoine Lévêque Laurent Clavier Michel Vasilivski Andreas Kaiser Marie-Minerve Loërat IEMN LIP6-UPMC Supported by: ANR WASABI Project W ireless system A nd S ystemC-AMS : B asic I nfrastructure Partners: STMicroelectronics Grenoble, Magillem Design Services A. M A. MAS ASSOUR OURI 1
MOTIVATION September 23-24, 2010 San Jose, California, USA AMS&RF SoC virtual prototyping and validation through the development of an industrial example: Wireless Video system (WVS). Inter-Vehicle Communication SoC Vehicle 1 SoC Vehicle 2 Wireless Optical MIPS RF RF MIPS Optical Camera DSP Transceiver Transceiver DSP Camera Channel RF domain Analog domain Digital domain Optical domain Heterogeneous system What is the efficient behavioral RF model to be used in this complex application? Which tool/programming language should be used to model & simulate this application ? A. M A. MAS ASSOUR OURI 2
OUTLINE September 23-24, 2010 San Jose, California, USA RF models and Modeling language Application & Simulation Platform Wireless Channel Model Simulation results Prospects A. A. M MAS ASSOUR OURI 3
OUTLINE September 23-24, 2010 San Jose, California, USA Modeling language: SystemC-AMS Application & Simulation Platform Wireless Channel Model Simulation results Summary A. A. M MAS ASSOUR OURI 4
BEHAVIORAL RF September 23-24, 2010 MODELS San Jose, California, USA Which Signal representation of RF Signal? Simulations can be done using either passband or complex baseband representation. Passband Pass-band simulations are more accurate However, they consume more resources and simulation time. Baseband Baseband models suppress the carrier frequency to trade some accuracy for a dramatic increase in execution speed (they run thousands of times faster than passband models) Baseband models allow a generic channel modeling Baseband behavioral models are used for RF devices and wireless channel A. M A. MAS ASSOUR OURI 5
MODELING LANGUAGE September 23-24, 2010 SYSTEMC-AMS San Jose, California, USA Why SystemC-AMS? According to previous works, it was established that SystemC-AMS is an efffecient tool to deal with the described application What is SystemC-AMS? Analog/Mixed-Signal (AMS) standard of the Open SystemC™ Initiative (OSCI) Open Source Model of Computation Martin Barnasconi “ SystemC AMS Extensions: Solving the Need for Speed, ” DAC -2010 AMS Working Group Chairman, Open SystemC Initiative, San Jose, CA USA 1. Electrical Linear Networks (ELN) : used to model continuous time behavior (current & voltage) 2. Linear Signal Flow (LSF) : used to model continuous time behavior 3. Timed Data Flow (TDF) : facilitates a very effecient simulation, as TDF models are processed at discret time points without using the discret-event kernel of SystemC TDF is the SystemC-AMS formalism used in this work A. M A. MAS ASSOUR OURI 6
OUTLINE September 23-24, 2010 San Jose, California, USA Modeling language: SystemC-AMS Application & Simulation Platform Wireless Channel Model Simulation results Summary A. A. M MAS ASSOUR OURI 7
APPLICATION & SIMULATION September 23-24, 2010 PLATFORM San Jose, California, USA STIMULUS STIMULUS o Positions, speeds CAR o QPSK modulation, Non-linearities, … Car-1 Car-2 Car-3 Car-4 Car-5 Wireless Channel o multipath Distances o Time-varying Speeds o Broadcasting characteristics Link 1 to 2 20 wireless time-varying channel Link 1 to 3 Wireless Channel Link 1 to 4 o Time and memory consuming Link 1 to 5 A. M A. MAS ASSOUR OURI 8
OUTLINE September 23-24, 2010 San Jose, California, USA Modeling language: SystemC-AMS Application & Simulation Platform Wireless Channel Model Simulation results Summary A. A. M MAS ASSOUR OURI 9
WIRELESS CHANNEL September 23-24, 2010 MODELING San Jose, California, USA What do we mean by wireless communication channel? Transmitter ( Tx ) Receiver ( Rx ) The transmission channel comprises antennas and all objects contributing or hampering propagation between source and destination nodes The propagation channel excludes the antennas and expresses all wave propagation phenomena between Tx and Rx Transmission channel is considered in this work !! A. M A. MAS ASSOUR OURI 10
WIRELESS CHANNEL September 23-24, 2010 MODELING San Jose, California, USA − ( ) − K 1 Receiver ( Rx ) k ∑ Transmitter ( Tx ) ~ = ⋅ χ ⋅ + LdB / 10 y ( t ) 10 g ( t ). x t n ( t ) k f = k 0 s − ( ) − K 1 ∑ k ~ = ⋅ χ ⋅ δ + LdB / 10 h ( t ) 10 g ( t ). t n ( t ) k f = k 0 s A. M A. MAS ASSOUR OURI 11
WIRELESS CHANNEL September 23-24, 2010 MODELING San Jose, California, USA Node-to-Node Link − ( ) − K 1 ∑ k ~ = ⋅ χ ⋅ δ + LdB / 10 h ( t ) 10 g ( t ). t n ( t ) k f = k 0 s − log normal − − ( ) K 1 ∑ k = ⋅ γ ⋅ ~ = δ n ( t ) h ( t ) g ( t ). t L 10 log d MPC k dB 10 Km f = shadowing k 0 s Path Loss Shadowing Small Scale Fading : Multipath AWGN Mean attenuation at a given Environment Reflection, diffraction, diffusion, refraction, ... Non-idealities of Simple Complex distance Antenna Simple Short time of simulation Memory Consuming Simple Short time of simulation Simulation time consuming Short time of simulation Time-varying multipath contribution will be detailed !! A. M A. MAS ASSOUR OURI 12
WIRELESS CHANNEL September 23-24, 2010 MODELING San Jose, California, USA Small Scale Fading Contribution : Multipath propagation Tapped Delay Line (TDL) − − K 1 k ∑ ~ = δ h ( t ) g ( t ). t o Uniformly spaced model MPC k f = o FIR filter (order K = number of paths) k 0 s o Coefficients are complex Gaussian variables 1 x ( t ) 1 f f s s ~ ~ ~ g 0 t ( ) g K − 1 t ( ) g 1 t ( ) Filtered Gaussian Noise o 2 independent Gaussian Variables (Box-Muller method) o Time-varying criteria: Doppler filter − − K 1 ∑ k ~ Real = y ( t ) g ( t ). x t k Gaussian f = k 0 s Random Variable ~ ~ ~ = + ⋅ I Q g ( t ) g ( t ) j g ( t ) k k k Real Gaussian Random Variable j Doppler Filter Design !! A. M A. MAS ASSOUR OURI 13
TIME-VARIANT CHANNEL FADING DOPPLER SPECTRUM September 23-24, 2010 San Jose, California, USA Doppler Shift? Motion of cars or scatterers produces Doppler θ k shifts of incoming received waves ( ) v Frequency shift ~ Doppler spread = + ⋅ θ = Tx Rx f f f cos , f λ k c d k d Time-varying aspect of the wireless channel is due to this physical phenomenon Which Doppler Spectrum for Mobile Communication? Jakes, Flat, Gaussian, Rounded, … Jakes Doppler Spectrum 1 = ≤ FIR Filter (High order) Spectrum S ( f ) , f f ( ) d π − 2 f 1 f f d d IIR Filter (Stability problem) Amplitude frequency ⇓ = Power Spectrum: « U Shape » H ( f ) S ( f ) Butterworth Modeling & Implementation o f d << f s Chebuchev Type-I/II o cut-off frequency is sharp Elleptic ( attenuation in the stop o Frequency-domain: ( simple, but all the channel coefficients band,No ripple in stop band ) must be generated in the beginning of the simulation) Time-domain: ( complex, but it has the real-time aspect of wireless communication ) A. M A. MAS ASSOUR OURI 14
Recommend
More recommend