#
# SPDX-FileCopyrightText: Copyright (c) 2021-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
"""Userful metrics for the Sionna library."""
import tensorflow as tf
from tensorflow.keras.metrics import Metric
from tensorflow.keras.losses import BinaryCrossentropy
[docs]class BitErrorRate(Metric):
"""BitErrorRate(name="bit_error_rate", **kwargs)
Computes the average bit error rate (BER) between two binary tensors.
This class implements a Keras metric for the bit error rate
between two tensors of bits.
Input
-----
b : tf.float32
A tensor of arbitrary shape filled with ones and
zeros.
b_hat : tf.float32
A tensor of the same shape as ``b`` filled with
ones and zeros.
Output
------
: tf.float32
A scalar, the BER.
"""
def __init__(self, name="bit_error_rate", **kwargs):
super().__init__(name, **kwargs)
self.ber = self.add_weight(name="ber",
initializer="zeros",
dtype=tf.float64)
self.counter = self.add_weight(name="counter",
initializer="zeros",
dtype=tf.float64)
def update_state(self, b, b_hat):
self.counter.assign_add(1)
self.ber.assign_add(compute_ber(b, b_hat))
def result(self):
#cast results of computer_ber for compatibility with tf.float32
return tf.cast(tf.math.divide_no_nan(self.ber, self.counter),
dtype=tf.float32)
def reset_state(self):
self.ber.assign(0.0)
self.counter.assign(0.0)
[docs]def compute_ber(b, b_hat):
"""Computes the bit error rate (BER) between two binary tensors.
Input
-----
b : tf.float32
A tensor of arbitrary shape filled with ones and
zeros.
b_hat : tf.float32
A tensor of the same shape as ``b`` filled with
ones and zeros.
Output
------
: tf.float64
A scalar, the BER.
"""
ber = tf.not_equal(b, b_hat)
ber = tf.cast(ber, tf.float64) # tf.float64 to suport large batch-sizes
return tf.reduce_mean(ber)
[docs]def compute_ser(s, s_hat):
"""Computes the symbol error rate (SER) between two integer tensors.
Input
-----
s : tf.int
A tensor of arbitrary shape filled with integers indicating
the symbol indices.
s_hat : tf.int
A tensor of the same shape as ``s`` filled with integers indicating
the estimated symbol indices.
Output
------
: tf.float64
A scalar, the SER.
"""
ser = tf.not_equal(s, s_hat)
ser = tf.cast(ser, tf.float64) # tf.float64 to suport large batch-sizes
return tf.reduce_mean(ser)
[docs]def compute_bler(b, b_hat):
"""Computes the block error rate (BLER) between two binary tensors.
A block error happens if at least one element of ``b`` and ``b_hat``
differ in one block. The BLER is evaluated over the last dimension of
the input, i. e., all elements of the last dimension are considered to
define a block.
This is also sometimes referred to as `word error rate` or `frame error
rate`.
Input
-----
b : tf.float32
A tensor of arbitrary shape filled with ones and
zeros.
b_hat : tf.float32
A tensor of the same shape as ``b`` filled with
ones and zeros.
Output
------
: tf.float64
A scalar, the BLER.
"""
bler = tf.reduce_any(tf.not_equal(b, b_hat), axis=-1)
bler = tf.cast(bler, tf.float64) # tf.float64 to suport large batch-sizes
return tf.reduce_mean(bler)
[docs]def count_errors(b, b_hat):
"""Counts the number of bit errors between two binary tensors.
Input
-----
b : tf.float32
A tensor of arbitrary shape filled with ones and
zeros.
b_hat : tf.float32
A tensor of the same shape as ``b`` filled with
ones and zeros.
Output
------
: tf.int64
A scalar, the number of bit errors.
"""
errors = tf.not_equal(b,b_hat)
errors = tf.cast(errors, tf.int64)
return tf.reduce_sum(errors)
[docs]def count_block_errors(b, b_hat):
"""Counts the number of block errors between two binary tensors.
A block error happens if at least one element of ``b`` and ``b_hat``
differ in one block. The BLER is evaluated over the last dimension of
the input, i. e., all elements of the last dimension are considered to
define a block.
This is also sometimes referred to as `word error rate` or `frame error
rate`.
Input
-----
b : tf.float32
A tensor of arbitrary shape filled with ones and
zeros.
b_hat : tf.float32
A tensor of the same shape as ``b`` filled with
ones and zeros.
Output
------
: tf.int64
A scalar, the number of block errors.
"""
errors = tf.reduce_any(tf.not_equal(b,b_hat), axis=-1)
errors = tf.cast(errors, tf.int64)
return tf.reduce_sum(errors)