I want to do a 2D elastic wave simulation on step geometry

Hello,

I am new to k-wave simulation, and I am trying to simulate the attenuation through media. First of all, for medium.alpha_coeff, what should the unit be? dB/MHz.cm or Np/MHz.cm? Secondly, how do I make sure the attenuation through the medium is correct? I placed two sensors, one near the source and the other 1.1mm away, and now by looking at sensor signals, I see there is an attenuation, but if I consider the peak-to-peak value of the sensor signal, it does not follow this relationship: P=P0*exp(-alpha*f*x), why? Here is my code:
% Define medium properties
medium.sound_speed = zeros(Nx, Ny);
medium.density = zeros(Nx, Ny);

medium.sound_speed(:,:) = 5555;
medium.density(:,:) = 2440;

% Define attenuation coefficients (Np/(MHz^y cm))
medium.alpha_coeff = zeros(Nx, Ny); % in dB/(MHz^y cm) but K-Wave expects Np/(MHz^y m)
medium.alpha_coeff(:,:) = 0.1; % low loss (converted to Np/(MHz^y m))
medium.alpha_power = 1.0; % frequency power
medium.alpha_mode = 'no_dispersion';

% Simulation time
cfl=0.3;
t_end = 16e-6; % [s]
sound_speed=5555;
dt = cfl * dx / sound_speed;
kgrid.makeTime(sound_speed, cfl,t_end)
% Tone burst source settings
sample_freq = 20e6; % [Hz]
signal_freq = 1e6; % [Hz]
num_cycles = 3;
signal=toneBurst(sample_freq,signal_freq,num_cycles,'Plot',true);

%Defining a single source point
source.p_mask = zeros(Nx, Ny);
source.p_mask(round(Nx/2), 1) = 1;

% define a time varying sinusoidal source
source_mag = 1e2; % Amplitude in Pascals
source.p=source_mag *signal;

% Initialize sensor mask with unique labels
sensor.mask = zeros(Nx, Ny);

% Sensor 1: label '1' in second row (just below source)
sensor.mask(round(Nx/2), 2) = 1;

% Sensor 2: label '2' in the row right after the glass layer
sensor.mask(round(Nx/2), 24) = 1;

sensor.record = {'p', 'p_final','I'};

% Run the simulation with memory-safe options
input_args = {'PMLInside', false, 'PMLSize', 20, 'PMLAlpha',3, 'PlotPML', true, 'DataCast', 'single', 'PlotLayout', true,'RecordMovie', true};
% input_args = {'DisplayMask', display_mask,'PMLInside', false, 'PMLSize', 20, 'PMLAlpha',2, 'PlotPML', true, 'DataCast', 'single', 'PlotLayout', true};
sensor_data = kspaceFirstOrder2D(kgrid, medium, source, sensor,input_args{:});

% Extract pressure signals
p_signals = sensor_data.p; % [2 x Nt]
Nt = length(kgrid.t_array);
t_micro = (0:Nt-1) * kgrid.dt * 1e6;

% Plot recorded signals
figure;
plot(t_micro, p_signals(1,:), 'b-', 'LineWidth', 1.5);
xlabel('Time (µs)');
ylabel('Pressure (Pa)');
legend('Sensor 1 (near source)');
title('Pressure1 Time‑Series Recorded by Sensors');
figure;
plot(t_micro, p_signals(2,:), 'r-', 'LineWidth', 1.5);
xlabel('Time (µs)');
ylabel('Pressure (Pa)');
legend('Sensor 2 (after 1.1mm)');
title('Pressure2 Time‑Series Recorded by Sensors');
grid on;
[row_sens, col_sens] = find(sensor.mask);
figure;
imagesc(x_mm, y_mm, medium.sound_speed'); axis image tight;
colormap(jet); colorbar;
xlabel('x (mm)'); ylabel('y (mm)');
title('Medium with Sensor and Source Locations');
hold on;

% Plot source in red
[row_src, col_src] = find(source.p_mask);
plot(x_mm(row_src), y_mm(col_src), 'rs', 'MarkerSize', 8, 'LineWidth', 1.5);

% Plot sensors in green
plot(x_mm(row_sens), y_mm(col_sens), 'go', 'MarkerSize', 8, 'LineWidth', 1.5);
legend('Source', 'Sensor');

k-Wave

A MATLAB toolbox for the time-domain
simulation of acoustic wave fields