3. Application of to RFI detection

The idea is to use the real time capabilities of the correlators to compute the vector .

Denote g(x₁,x₂) as the quantized product between level x₁ and level x₂; and l_i the quantization levels. When and , the value of the quantized product g(x₁, x₂) is given by m_i,j. Table 1 (click here) gives an example of a quantization scheme. By the means of g(x₁,x₂), a set of Q bivariate (L_k=2) non-linear functions F_k can be obtained, namely:

 

By taking the mean of these functions, the generalized moments, w_k, given by Eq. (1 (click here)) and the vector given by Eq. (2 (click here)) are generated. In fact, the proposed vector corresponds to Q time-lags of the quantized correlator and it is instantaneously obtained through the correlator buffers after N clock cycles. Thus, only the matrix operations of Eq. (3 (click here)) need to be externally calculated in real time.

These matrix operations use the knowledge of the reference vector and the covariance matrix . In practice, the latter are estimated from the observations. This point is addressed in the following section.