Added entropy, average, stddev, variance, bytes ratio calculation

3 files changed, 43 insertions, 4 deletions

diff --git a/src/include/ndpi_api.h b/src/include/ndpi_api.h
index 03f21d4cd..c6d17a4b5 100644
--- a/src/include/ndpi_api.h
+++ b/src/include/ndpi_api.h
@@ -899,8 +899,11 @@ extern "C" {
   float ndpi_data_average(struct ndpi_analyze_struct *s);
   float ndpi_data_window_average(struct ndpi_analyze_struct *s);
   
-  float ndpi_entropy(struct ndpi_analyze_struct *s);
-
+  float ndpi_data_entropy(struct ndpi_analyze_struct *s);
+  float ndpi_data_variance(struct ndpi_analyze_struct *s);
+  float ndpi_data_stddev(struct ndpi_analyze_struct *s);
+  float ndpi_data_ratio(u_int32_t sent, u_int32_t rcvd);
+    
   void ndpi_data_print_window_values(struct ndpi_analyze_struct *s); /* debug */
 #ifdef __cplusplus
 }
diff --git a/src/include/ndpi_typedefs.h b/src/include/ndpi_typedefs.h
index 2aacf847a..0db1ccf91 100644
--- a/src/include/ndpi_typedefs.h
+++ b/src/include/ndpi_typedefs.h
@@ -1369,6 +1369,11 @@ struct ndpi_analyze_struct {
   u_int32_t *values;
   u_int32_t sum_total, num_data_entries, next_value_insert_index;
   u_int16_t num_values_array_len /* lenght of the values array */;
+
+  struct {
+    /* https://www.johndcook.com/blog/standard_deviation/ */
+    float mu, q;
+  } stddev;
 };
 
 #define DEFAULT_SERIES_LEN  64
diff --git a/src/lib/ndpi_analyze.c b/src/lib/ndpi_analyze.c
index 502c2d858..ce3168165 100644
--- a/src/lib/ndpi_analyze.c
+++ b/src/lib/ndpi_analyze.c
@@ -72,20 +72,42 @@ void ndpi_free_data_analysis(struct ndpi_analyze_struct *d) {
   Add a new point to analyze
  */
 void ndpi_data_add_value(struct ndpi_analyze_struct *s, const u_int32_t value) {
+  float tmp_mu;
+
   s->sum_total += value, s->num_data_entries++, s->values[s->next_value_insert_index] = value;
+
   if(++s->next_value_insert_index == s->num_values_array_len)
     s->next_value_insert_index = 0;
+
+  /* Update stddev */
+  tmp_mu = s->stddev.mu;
+  s->stddev.mu = ((s->stddev.mu * (s->num_data_entries - 1)) + value) / s->num_data_entries;
+  s->stddev.q = s->stddev.q + (value - tmp_mu)*(value - s->stddev.mu);    
 }
 
 /* ********************************************************************************* */
 
-/* Compute the average on all value */
+/* Compute the average on all values */
 float ndpi_data_average(struct ndpi_analyze_struct *s) {
   return((float)s->sum_total / (float)s->num_data_entries);
 }
 
 /* ********************************************************************************* */
 
+/* Compute the variance on all values */
+float ndpi_data_variance(struct ndpi_analyze_struct *s) {
+  return(s->num_data_entries ? (s->stddev.q / s->num_data_entries) : 0);
+}
+
+/* ********************************************************************************* */
+
+/* Compute the standard deviation on all values */
+float ndpi_data_stddev(struct ndpi_analyze_struct *s) {
+  return(sqrt(ndpi_data_variance(s)));
+}
+
+/* ********************************************************************************* */
+
 /* Compute the average only on the sliding window */
 float ndpi_data_window_average(struct ndpi_analyze_struct *s) {
   float   sum = 0.0;
@@ -102,7 +124,7 @@ float ndpi_data_window_average(struct ndpi_analyze_struct *s) {
 /*
   Compute entropy on the last sliding window values
 */
-float ndpi_entropy(struct ndpi_analyze_struct *s) {
+float ndpi_data_entropy(struct ndpi_analyze_struct *s) {
   int i;
   float sum = 0.0, total = 0.0;
 
@@ -129,3 +151,12 @@ void ndpi_data_print_window_values(struct ndpi_analyze_struct *s) {
 
   printf("\n");
 }
+
+/* ********************************************************************************* */
+
+float ndpi_data_ratio(u_int32_t sent, u_int32_t rcvd) {
+  int64_t s = (int64_t)sent +  (int64_t)rcvd;
+  int64_t d =  (int64_t)sent -  (int64_t)rcvd;
+
+  return((s == 0) ? 0 : ((float)d)/((float)s));
+}