Input/Output-to-State Stability and Output-to-State Stability of Direct-Form Digital Filters with Finite Wordlength Nonlinearity

1252 Words3 Pages

This paper proposes a new criterion for the input/output-to-state stability (IOSS) of interfered direct-form digital filters with finite wordlength nonlinearity based on an augmented Lyapunov function. Without external interference, the output-to-state stability (OSS) and asymptotic stability of direct-form digital filters with finite wordlength nonlinearity are also guaranteed under the proposed criterion. This criterion is expressed by linear matrix inequalities (LMIs). A numerical example demonstrates the effectiveness of the proposed criterion.
Key Words: input/output-to-state stability (IOSS); output-to-state stability (OSS); asymptotic stability; digital filter; finite wordlength effect

1 Introduction
When we design and implement digital filters, we use fixed-point arithmetic, which generates overflow and quantization nonlinearities. These nonlinearities can cause the digital filters potentially to produce zero-input limit cycles if the coefficients of the digital filters are not selected appropriately [?]. Zero-input limit cycles are an unstable behavior and should be avoided in digital filter design [?]. Thus, the range determination of the coefficients of digital filters, where these limit cycles should be avoided, is very important. Much attention has been focused on the criteria for the removal of limit cycles in digital filters, including overflow nonlinearity [?, ?, ?, ?, ?, ?, ?, ?, ?, ?]. When we implement a high-order large scaled digital filter using hardware and digital computers, we usually break it down into several low-order digital filters before implementation. In this case, mutual interferences inevitably occur between these low-order filters, and this leads to malfunction as well as poor perfo...

... middle of paper ...

...hich guarantee the IOSS when and the OSS and asymptotic stability when in this example. Let . Figure 1 shows that the state variables are bounded around the origin by the IOSS property when , where is a white Gaussian random sequence with mean and variance . Figure 2 shows that the state variables converge to the origin when .

Figure 1: Phase plot when

Figure 2: Phase plot when

5 Conclusion
In this paper, we have proposed a new LMI based criterion for the IOSS of interfered direct-form digital filters with saturation overflow nonlinearity. Based on the augmented Lyapunov function, the criterion also guarantees the OSS and asymptotic stability of direct-form digital filters without external interference under the additional LMI condition. The criterion is expressed by LMIs. An example was provided to demonstrate the proposed criterion.

More about Input/Output-to-State Stability and Output-to-State Stability of Direct-Form Digital Filters with Finite Wordlength Nonlinearity

Open Document