Denoising very noisy and large oscilloscope data

Question

I have UHF data with 100 million data points and a broad band of background noise (about 5 mV). Is there a way to filter that noise out? Due to there being so many data points every filter function I try to implement just smoothes the data lines including the noise. The result is that you cannot see any filter effect if you are zoomed out.

Filter Example:

b, a = scipy.signal.butter(15, 0.1)
y = scipy.signal.filtfilt(b,a, df[1].values)
df2 = pd.DataFrame(y, index=df.index)

Unfortunately I do not know how to provide you with the high quality data that is shown to me while plt.show() . There I can zoom to every data point. When I save the plot to an image, quality drops to 600 or so dpi.

UHF file: https://ufile.io/hhkeg

Code:

import matplotlib.pyplot as plt
import pandas as pd
import readTrc
from scipy import signal

datX, datY, m = readTrc.readTrc('C220180104_ch2_UHF00000.trc')
#100mil data points
srx, sry = pd.Series(datX * 1000), pd.Series(datY * 1000)
df = pd.concat([srx, sry], axis = 1)
df.set_index(0, inplace = True)

b, a = signal.butter(15, 0.1)
y = signal.filtfilt(b,a, df[1].values)
df2 = pd.DataFrame(y, index=df.index)

#Plot Impulse
if max(datX) < 0.01:
    df2.plot(grid = 1,
            linewidth = 0.5,
            figsize = (9,5),
            legend = False,
            xlim = (df.index[0], df.index[-1]))
else:
    df2.plot(grid = 1,
            linewidth = 1,
            figsize = (9,5),
            legend = False,
            xlim = (df.index[0], df.index[-1]))

plt.xlabel('Zeit in ms')
plt.ylabel('UHF-Signal in mV')

##plt.savefig('UHF_plot.png', dpi = 600)
plt.show()

print('done')

Without Filter:

With Filter:

There is really not much use out of that.