main.py

from db.utils import get_connection
from data.read import read_data
import numpy as np
from scipy import stats
from utils.distance import haversine
import pandas as pd

def prepare_data() -> pd.DataFrame:
    conn = get_connection("exported_database.sql")
    df = read_data(conn)
    return df

def calc_distances(df: pd.DataFrame):
    base_station_loc = df.iloc[0][['latitude', 'longitude']]
    df['distance'] = df.apply(lambda row: haversine(base_station_loc['latitude'], base_station_loc['longitude'], row['latitude'], row['longitude']), axis=1)

if __name__ == "__main__":
    df = prepare_data()
    # df = pd.read_csv("drive_test.csv")
    calc_distances(df)

    df = df[df['distance'] > 0.0]
    
    # Pr = P0 - 10 * beta * log10(d/d0) + σ²

    Pr = df['rsrp'].to_numpy()
    d = df['distance'].to_numpy()

    d0 = 1.0  # Reference distance (d0) in km, to match the unit of our calculated distances
    X = 10 * np.log10(d / d0)

    # Perform linear regression
    slope, intercept, _, _, _ = stats.linregress(X, Pr)

    beta_estimate = -slope
    P0_estimate = intercept

    # Estimate the noise standard deviation (sigma) by calculating the residuals
    residuals = Pr - (P0_estimate - beta_estimate * X)
    sigma_squared_estimate = np.var(residuals)
    sigma_estimate = np.sqrt(sigma_squared_estimate)

    print("Estimated Parameters:\n")
    print("Power (P0): {:.2f} dBm".format(P0_estimate))
    print("Path Loss Exponent (beta): {:.2f}".format(beta_estimate))
    print("Gaussian Noise Standard Deviation: {:.2f}".format(sigma_estimate))