Method of managing a traffic load

1. A method of managing a traffic load of a network node for receiving a packet traffic aggregated from a multitude of application streams in a packet switched network, comprising:

a) estimating a granularity of the packet traffic as a maximum number of the application streams which fit into a transmission capacity B of the network node;

b) calculating a drop probability P d based on the estimated granularity and the traffic load of the network node; and

c) providing the calculated drop probability P d for a congestion control, wherein the step a) comprises,

determining a capacity utilisation ratio x as a time-averaged ratio of the traffic load of the network node to the transmission capacity B, whereby 0≦x≦1 and the time averaging is on a first time scale;

determining a time-averaged value m 1 of the capacity utilisation ratio x and a time-averaged value m 2 of the square x 2 of the capacity utilisation ratio x, whereby the time averaging is on a second time scale which is longer than the first time scale; and

calculating N=m 1 /(m 2 −(m 1 ) 2 ) as the said estimated maximum number of the application streams which fit into the transmission capacity B of the network node, wherein

m 1 , m 2 , N and x are real numbers.

2. The method of claim 1 , wherein the first time scale is comparable to a buffer holding time of the network node.

3. The method of claim 1 , wherein the first time scale is in a range from 1 to 10 ms.

4. The method of claim 1 , wherein the second time scale is greater than 1 s.

5. The method of claim 1 , wherein the second time scale is at least 100 times greater than the first time scale.

6. The method of claim 1 , wherein the step b) comprises:

assuming that a probability P(k) that a packet traffic with an average of A application streams consists of k application streams follows a Poisson distribution with P(k)=A k e −A /k!;

calculating an overflow probability P ov as the sum of the probabilities P(k) for k>N whereby N is the maximum number of the application streams which fit into the transmission capacity B of the network node; and

assuming the drop probability P d to be less than the overflow probability P ov , wherein

e, A and k are real numbers.

7. The method of claim 1 , wherein the method further comprises:

dropping and/or marking a packet of the packet traffic which is being received in accordance with the calculated drop probability P d .

8. The method of claim 1 , wherein the method further comprises:

initiating a congestion notification in accordance with the calculated drop probability P d .

9. The method of claim 8 , wherein the method further comprises:

triggered by the congestion notification, reducing a rate of the packet traffic.

10. A network node for receiving a packet traffic aggregated from a multitude of application streams in a packet switched network, the network node comprising:

a control unit configured to estimate a granularity of the packet traffic as a maximum number of the application streams which fit into a transmission capacity B of the network node;

calculate a drop probability P d based on the estimated granularity and the traffic load of the network node; and provide the calculated drop probability P d for a congestion control, wherein the control unit, for the said estimation of a granularity of the packet traffic, is further configured to,

determine a capacity utilisation ratio x as a time-averaged ratio of the traffic load of the network node to the transmission capacity B, whereby 0≦x≦1 and the time averaging is on a first time scale;

determine a time-averaged value m 1 of the capacity utilisation ratio x and a time-averaged value m 2 of the square x 2 of the capacity utilisation ratio x, whereby the time averaging is on a second time scale which is longer than the first time scale; and

calculate N=m 1 /(m 2 −(m−m 1 ) 2 ) as the said estimated maximum number of the application streams which fit into the transmission capacity B of the network node, wherein m 1 , m 2 , N and x are real numbers.

11. A computer program product comprising:

a non-transitory computer readable medium having computer executable code for managing a traffic load of a network node for receiving a packet traffic aggregated from a multitude of application streams in a packet switched network, the computer executable code when executed causing a processor of the network node to perform:

a) estimating a granularity of the packet traffic as a maximum number of the application streams which fit into a transmission capacity B of the network node;

b) calculating a drop probability P d based on the estimated granularity and the traffic load of the network node; and

c) providing the calculated drop probability P d for a congestion control, wherein the step a) comprises,

determining a capacity utilisation ratio x as a time-averaged ratio of the traffic load of the network node to the transmission capacity B, whereby 0≦x≦1 and the time averaging is on a first time scale;

determining a time-averaged value m 1 of the capacity utilisation ratio x and a time-averaged value m 2 of the square x 2 of the capacity utilisation ratio x, whereby the time averaging is on a second time scale which is longer than the first time scale; and

calculating

N

=

m

1

(

m

2

-

(

m

-

m

1

)

2

)

as the said estimated maximum number of the application streams which fit into the transmission capacity B of the network node, wherein m 1 , m 2 , N and x are real numbers.