Mean Opinion Score (MOS)

The metric used for estimating voice quality is a Mean Opinion Score (MOS) based on the ITU-T E-model (► ITU-T Recommendation G.107). The inputs are network statistics such as speech codec usage, network delay, jitter, and packet loss. The modified E-model outputs an R-value, which is straightforwardly converted to a MOS value.

The MOS scale is laid out in the following table:

MOS	Quality	Impairment
5	Excellent	Imperceptible
4	Good	Perceptible
3	Fair	Annoying
2	Poor	Very annoying
1	Bad	Impossible to communicate

Calculation of MOS in Paragon Active Assurance

The algorithm for calculating MOS in the VoIP task is given below. It follows the ITU-T E-model.

For SIP, the algorithm is the same, but the average delay (davg) is assumed to be 5 ms in this case.

The function takes the following arguments:

• loss = packet loss in %

• davg = average delay

• dmin = minimum delay

• dmax = maximum delay

• ie = equipment impairment factor

• bpl = packet-loss robustness factor (codec-specific)

float Stats::calc_mos(float loss, float davg, float dmin, float dmax, float ie, float bpl)
{
    float mos, r, deff;


    deff = davg + 2 * (dmax - dmin) + 10;
    if (deff < 160)
    {
        r = 93.2 - deff / 40;
    }
    else
    {
        r = 93.2 - (deff - 120) / 12;
    }
    r -= ie + (95 - ie) * loss / (loss + bpl);
    if (r < 0)
    {
        mos = 1;
    }
    else
    {
        mos = 1 + 0.035 * r + 0.000007 * r * (r - 60) * (100 - r);
    }
    return mos;
}