✅作者简介:热爱科研的Matlab仿真开发者,修心和技术同步精进,代码获取、论文复现及科研仿真合作可私信。
个人主页:Matlab科研工作室
个人信条:格物致知。
更多Matlab完整代码及仿真定制内容点击
智能优化算法 神经网络预测 雷达通信 无线传感器 电力系统
信号处理 图像处理 路径规划 元胞自动机 无人机
智能传感技术预计将成为智能城市和近期环境中的普遍技术。由于集成设备尺寸缩小,同时又保持其计算能力,这些服务正在提高其能力,这些设备可以运行各种机器学习算法,并在各种数据处理任务中实现高性能。一种可用于智能传感的有吸引力的传感器方式是声传感器,它可以在保持中等能耗的情况下传递高信息量的数据。不幸的是,目前无线传感器网络的能量预算通常不足以满足标准麦克风的要求。因此,需要提高各层次----传感、信号处理和通信----的能源效率,以便将无线智能声学传感器推向市场。为了实现这一目标,本文介绍了一种用于无线声传感器网络的能量消耗模型,其目的是帮助开发新技术,以提高智能无线声传感器的能源效率。该模型为智能无线声传感器的定制设计提供了第一步探索,也可用于比较不同协议的能耗。也可用于比较不同协议的能耗.也可用于比较不同协议的能耗.
% ----------------------------------------------------------------------
% Script which estimates the energy consumption, battery lifetime and
% needed computational resource of a smart audio sensor chain.
% The chain includes sensing, processing and wireless communication.
%
% The model is fully described in the following paper:
% G. Dekkers, F. Rosas, S. Lauwereins, S. Rajendran, S. Pollin, B. Vanrumste, T. van Waterschoot, M. Verhelst and Peter Karsmakers,
% 揂 multi-layered energy consumption model for smart wireless acoustic sensor networks,?KU Leuven, Tech. Rep., December 2018.
% Please refer to this paper when using this model for your research.
%
% Authors:
% Gert Dekkers, KU Leuven
% Fernando Rosas, Imperial College London
% ----------------------------------------------------------------------
clc; clear;
addpath(genpath(pwd))
%% Code for acquiring energy consumption for a single estimate (network node)
verbose = 1; %0: show final output, 1: show all
% -- Init system -- %
% define setup
gen.method = 'default'; % name of the general config file
sens.method = 'default'; % name of sensing config file
% proc{1}.method = 'ADC';
proc{1}.method = 'FE/framelogMel'; % name of the first processing chain (typically FE). if no FE => 'FE/ADC'
proc{2}.method = 'NN/CNN_1D_example'; % name of second processing chain (typically classifier) or remove line
comm.method = 'default'; % name of communication config file
% get general param
gen = general_loadparam(gen.method,[]); % get general parameters (fs, 礐, ..)
% -- Sensing layer -- %
if verbose, disp('%%%% Sensing %%%%'); end;
sens.conf = sensing_loadparam(sens.method, [], gen); % get sensing parameters
[sens.cons, sens.os] = sensing_consumption(sens.conf, gen); % [mJ]
% -- Processing layer -- %
input_shape = sens.os; prev_chain = []; E_proc = []; name_proc = cell(0); %init as empty at first run
for p=1:length(proc) % for each processing chain
if verbose, disp(['%%%% Processing layer: ' proc{p}.method ' %%%%']); end;
% get values
proc{p}.conf = proc_loadparam(proc{p}.method, [], input_shape, gen); % get params
[proc{p}.os, proc{p}.ops, proc{p}.par] = proc_info(proc{p}.conf, gen, input_shape,verbose); % get output shape, ops and nr. parameters
[proc{p}.ma, proc{p}.ms_o, proc{p}.ms_p] = memo_acc_stor(proc{p}.os, proc{p}.par, proc{p}.ops, proc{p}.conf, prev_chain, gen,verbose); % get memory used in storing ops/params
[proc{p}.cons.all,proc{p}.cons.op,proc{p}.cons.ms_o,proc{p}.cons.ms_p,proc{p}.cons.ma] = bits_to_energy(proc{p}.ma, proc{p}.ms_o, proc{p}.ms_p, proc{p}.ops, gen); % energy consumption
% keep from previous chain
input_shape = proc{p}.os(end,:);
prev_chain = proc{p}.conf;
end
% -- Communication layer -- %
if verbose, disp('%%%% Communication %%%%'); end;
comm.conf = comm_loadparam(comm.method,[],gen); % get comm params
comm.conf.N_T = prod(input_shape)*gen.S; % informative bits to communicate
comm.conf.N_R = 0; % informative bits to receive
comm.cons = sum(comm_consumption(comm.conf,sens.conf,gen));
% -- Final output -- %
fc = prep_chain_info(sens,proc,comm,gen); % get everything in proper format to plot/print
print_chain_info(fc) % Print energy distribution of the full sensor network
plot_energy_proc(proc) % Plot energy distribution of particular layers in the processing chain
print_mem_cycle(proc,gen) % Print needed cycli/s and mem storage
G. Dekkers, F. Rosas, S. Lauwereins, S. Rajendran, S. Pollin, B. Vanrumste, T. van Waterschoot, M. Verhelst and P. Karsmakers, “A multi-layered energy consumption model for smart wireless acoustic sensor networks,” KU Leuven, Tech. Rep., December 2018.
2.1 bp时序、回归预测和分类
2.2 ENS声神经网络时序、回归预测和分类
2.3 SVM/CNN-SVM/LSSVM/RVM支持向量机系列时序、回归预测和分类
2.4 CNN/TCN卷积神经网络系列时序、回归预测和分类
2.7 ELMAN递归神经网络时序、回归\预测和分类
2.9 RBF径向基神经网络时序、回归预测和分类