Signal Processing I(23ECE112) Invertible and Non-invertible systems • A system is said to be invertible if distinct input signal produces a distinct output signal. • For an invertible system, every unique input maps to a unique output. • Hence, if we are given the output of this system, we can find the input signal that caused this output provided we know the behaviour of the system. In other words, we can find the inverse of the output. • If this is not possible, the system is said to be non-invertible Systems Invertible Non-invertible Examples of invertible and non invertible systems • Invertible systems: ๐ฆ ๐ก = 5๐ฅ ๐ก , ๐ฆ ๐ก = ๐ฅ ๐ก − ๐ก0 , ๐ฆ ๐ก = ๐ฅ 3๐ก • ๐ฆ[๐] = ๐ฅ[๐ − ๐0 ], ๐ฆ ๐ = σ๐๐=−∞ ๐ฅ[๐] • Non-invertible systems: ๐ฆ(๐ก) = ๐ฅ 2 (๐ก) , ๐ฆ ๐ก = 0, ๐ฆ ๐ก = |๐ฅ(๐ก)| • ๐ฆ[๐] = ๐ฅ[2๐],๐ฆ[๐] = |๐ฅ[๐]| Practice Problems • Check whether these systems are invertible or non-invertible ๐ฆ(๐ก) = sin(๐ฅ ๐ก ) ๐ฆ ๐ก =๐ฅ ๐ก +2 ๐ฆ ๐ก =แ ๐ฅ ๐ก 5 ๐๐ ๐ฅ(๐ก) ≤ 5 ๐๐กโ๐๐๐ค๐๐ ๐ ๐ฆ ๐ก = ๐ฅ 3๐ก + 2 ๐ฆ[๐] = ๐ฅ[2๐ − 3] ๐ฆ[๐] = cos(๐ฅ[๐]) ๐ฆ[๐] = ๐๐ฅ[๐] + ๐ ๐ฆ ๐ = 0.25๐ฅ[๐] Practice Problems • Check whether these systems are invertible or non-invertible ๐ฆ(๐ก) = sin(๐ฅ ๐ก )-ni ๐ฆ ๐ก =๐ฅ ๐ก +2 ๐ฅ ๐ก ๐ฆ ๐ก =แ 5 ๐ฆ ๐ก = ๐ฅ 3๐ก + 2 -i ๐ฆ[๐] = ๐ฅ[2๐ − 3]-ni ๐ฆ[๐] = cos(๐ฅ[๐])-ni ๐ฆ[๐] = ๐๐ฅ[๐] + ๐-i ๐ฆ ๐ = 0.25๐ฅ[๐]-i ๐๐ ๐ฅ(๐ก) ≤ 5 ๐๐กโ๐๐๐ค๐๐ ๐ Inverse of system • If a system is invertible, we can find an inverse system that performs the reverse action of the system. 1 • ๐ฆ ๐ก = 5๐ฅ ๐ก has inverse system ๐ฆ ๐ก = 5 ๐ฅ ๐ก • ๐ฆ ๐ก = ๐ฅ ๐ก − ๐ก0 has inverse system as ๐ฆ ๐ก = ๐ฅ ๐ก + ๐ก0 • If we connect a system in cascade with its own inverse system, we get back the input signal x(t) Inverse of a system • We can obtain the inverse of a system for discrete time systems also. • We call a system as identity system if the output is the same as the input. ๐ฆ ๐ก = ๐ฅ(๐ก) or ๐ฆ[๐] = ๐ฅ[๐] • The cascade connection of an invertible system with its inverse produces an identity system. Practice Problem • Check the inverse systems for those systems which are invertible ๐ฆ(๐ก) = sin(๐ฅ ๐ก ) ๐ฆ ๐ก =๐ฅ ๐ก +2 ๐ฆ ๐ก =แ ๐ฅ ๐ก 5 ๐๐ ๐ฅ(๐ก) ≤ 5 ๐๐กโ๐๐๐ค๐๐ ๐ ๐ฆ ๐ก = ๐ฅ 3๐ก + 2 ๐ฆ[๐] = ๐ฅ[2๐ − 3] ๐ฆ[๐] = cos(๐ฅ[๐]) ๐ฆ[๐] = ๐๐ฅ[๐] + ๐ ๐ฆ ๐ = 0.25๐ฅ[๐] Practice Problem • Check the inverse systems for those systems which are invertible ๐ฆ(๐ก) = sin(๐ฅ ๐ก ) - ni ๐ฆ ๐ก =๐ฅ ๐ก +2 x(t)=Y(t)-2 ๐ฆ ๐ก =เต ๐ฆ ๐ก = ๐ฅ 3๐ก + 2 x(t)=y(t/3 -2/3) ๐ฆ[๐] = ๐ฅ[2๐ − 3] -ni ๐ฆ[๐] = cos(๐ฅ[๐]) -NI ๐ฆ[๐] = ๐๐ฅ[๐] + ๐ –x(n)=[y(n)-c]/m ๐ฆ ๐ = 0.25๐ฅ[๐] x(n)=1/0.25 y(n) ๐ฅ ๐ก 5 ๐๐ ๐ฅ(๐ก) ≤ 5 ๐๐ผ ๐๐กโ๐๐๐ค๐๐ ๐ Causality • A system is called a causal system if its output value at the present time instance depend only on the input values at present and previous time instances, not at future time instances. • Causal systems are also called as non-anticipative systems. • Example: ๐ฆ(๐ก) = (t − 1)๐ฅ ๐ก , ๐ฆ ๐ = 4๐ฅ[๐], ๐ฆ ๐ก = ๐ฅ(๐ก − 3), ๐ฆ ๐ = ๐ฅ[๐ − 4], ๐ฆ ๐ก = log(๐ฅ(๐ก)), ๐ฆ ๐ = ๐ ๐ฅ[๐] • If the output of a system depends on future time instances of input, then we say that the system is noncausal. • Example: ๐ฆ(๐ก) = ๐ฅ 3๐ก , ๐ฆ ๐ = ๐ฅ[2๐], ๐ฆ ๐ก = ๐ฅ(๐ก + 5), ๐ฆ ๐ = ๐ฅ[๐ + 5] Practice Problem • Check if the following systems are causal or not ๐ฆ(๐ก) = ๐ฅ 0.25๐ก ๐ฆ ๐ก = ๐ฅ 2๐ก + 3 ๐ฆ ๐ก =x t +5 ๐ฆ ๐ = ๐ฅ ๐ + ๐ฅ[๐ + 2] ๐ฆ[๐] = ๐ฅ[5๐] ๐ฆ[๐] = ๐ฅ 2 [๐] Practice Problem • Check if the following systems are causal or not ๐ฆ(๐ก) = ๐ฅ 0.25๐ก -no ๐ฆ ๐ก = ๐ฅ 2๐ก + 3 -no ๐ฆ ๐ก = x t + 5 -causal ๐ฆ ๐ = ๐ฅ ๐ + ๐ฅ[๐ + 2]- no ๐ฆ[๐] = ๐ฅ[5๐] - no ๐ฆ[๐] = ๐ฅ 2 [๐] - causal Some observations • If a system is memoryless, then it is definitely a causal system. • Note that the converse is not true. i.e if a system is given as causal then it need not be memoryless. • Non-causal systems cannot be realized in practice if the future values of input to the system are not known. • Are the systems ๐ฆ ๐ก = ๐ฅ −๐ก and ๐ฆ ๐ก = ๐ฅ ๐ก + sin(๐ก + 2) causal? Time invariance • A system is time invariant if system characteristics does not change over time. • If a time invariant system produces an output y(t) for an input x(t) applied today, it will produce the same output y(t) when the same input signal x(t) is applied tomorrow or some other time. • This circuit is time invariant system if the value of the resistors do not change with respect to time. • If the resistor values change over time, then the given system is a time variant system Time invariance • Let the input to a continuous time system be x(t) and the corresponding output be given as y(t). • If the system is time invariant, then a shifted input ๐ฅ ๐ก − ๐ก0 will produce ๐ฆ ๐ก − ๐ก0 . i.e the output is the same signal but shifted by the same amount as the input signal. • Similarly, If the input to a discrete time system is x[n] with a corresponding output y[n], then for time invariant system a shifted input ๐ฅ ๐ − ๐0 will produce an output ๐ฆ ๐ − ๐0 Examples of time invariant systems • Check if the system ๐ฆ ๐ก = log(๐ฅ(๐ก)) is time invariant or not? • ๐ฆ ๐ก = log(๐ฅ(๐ก)), Now apply a new signal ๐ฅ1 ๐ก = ๐ฅ ๐ก − ๐ก0 . Then corresponding output ๐ฆ1 ๐ก is given by • ๐ฆ1 ๐ก = log ๐ฅ1 ๐ก = log(๐ฅ ๐ก − ๐ก0 ) • If ๐ฆ1 ๐ก = ๐ฆ(๐ก − ๐ก0 ) then the system is time invariant. • ๐ฆ ๐ก − ๐ก0 = log ๐ฅ ๐ก − ๐ก0 = ๐ฆ1 ๐ก . Therefore, the system is time invariant. Practice problem • Check if the system ๐ฆ ๐ = ๐ ๐ฅ[๐] is time invariant or not? • ๐ฆ ๐ = ๐ ๐ฅ[๐], Now apply a new signal ๐ฅ1 [๐] = ๐ฅ ๐ − ๐0 . Then corresponding output ๐ฆ1 [๐] is given by • ๐ฆ1 ๐ = ๐ ๐ฅ1 ๐ = ๐ ๐ฅ ๐ − ๐0 • If ๐ฆ1 [๐] = ๐ฆ ๐ − ๐0 then the system is time invariant. • ๐ฆ ๐ − ๐0 = ๐ − ๐0 ๐ฅ ๐ − ๐0 = ๐ ๐ฅ ๐ − ๐0 - ๐0 ๐ฅ ๐ − ๐0 ≠ ๐ฆ1 ๐ . Therefore the system is time variant. Practice problem • Check if the system ๐ฆ ๐ก = ๐ฅ(2๐ก) is time invariant or not? • ๐ฆ ๐ก = ๐ฅ(2๐ก), Now apply a new signal ๐ฅ1 ๐ก = ๐ฅ ๐ก − ๐ก0 . Then corresponding output ๐ฆ1 ๐ก is given by • ๐ฆ1 ๐ก = ๐ฅ1 2๐ก = ๐ฅ(2๐ก − ๐ก0 ) • If ๐ฆ1 ๐ก = ๐ฆ(๐ก − ๐ก0 ) then the system is time invariant. • ๐ฆ ๐ก − ๐ก0 = ๐ฅ 2 ๐ก − ๐ก0 = ๐ฅ(2๐ก − 2๐ก0 ) ≠ ๐ฆ1 ๐ก . Therefore, the system is time variant Systems Time invariant Time variant Practice problem • Check if the following systems are time invariant or not ๐ฆ ๐ก = 2๐ก ๐ฅ ๐ก ๐ก ๐ฆ ๐ก =๐ฅ 2 ๐ฆ ๐ก = 3x t − 3 ๐ฆ ๐ = 2๐ฅ ๐ ๐ฆ[๐] = ๐ฅ[2๐ − 3] ๐ฆ[๐] = ๐ ๐ ๐ฅ[๐] Practice problem • Check if the following systems are time invariant or not ๐ฆ ๐ก = 2๐ก ๐ฅ ๐ก - no ๐ฆ ๐ก =๐ฅ ๐ก 2 -no ๐ฆ ๐ก = 3x t − 3 -yes ๐ฆ ๐ = 2๐ฅ ๐ -yes ๐ฆ[๐] = ๐ฅ[2๐ − 3]-no ๐ฆ[๐] = ๐ ๐ ๐ฅ[๐]-no Homogeneity Property • Let the input to a continuous time system be x(t) and the output signal be y(t). • If the system is homogenous, then a new input ๐ฅ1 ๐ก = ๐ ๐ฅ ๐ก will produce a new output given by ๐ฆ1 ๐ก = ๐ ๐ฆ ๐ก where a is a constant. • In other words, if your input is scaled by a constant value then the output is also scaled by the same constant. • Example: ๐ฆ ๐ก = 5๐ฅ ๐ก • If we apply a new input ๐ฅ1 ๐ก = ๐ ๐ฅ ๐ก then the output will be ๐ฆ1 ๐ก = 5 ๐ ๐ฅ ๐ก Therefore the system is homogenous =๐ 5๐ฅ ๐ก = ๐ ๐ฆ(๐ก). Examples of homogenous systems • ๐ฆ[๐] = ๐ฅ[๐ − ๐0 ] • If we apply a new input ๐ฅ1 [๐] = ๐ ๐ฅ[๐] then the output will be ๐ฆ1 [๐] = ๐ฅ1 [๐ − ๐0 ] = ๐ ๐ฅ ๐ − ๐0 = ๐ ๐ฆ[๐]. Therefore, the system is homogenous • ๐ฆ(๐ก) = ๐๐๐(๐ฅ ๐ก ) • If we apply a new input ๐ฅ1 ๐ก = ๐ ๐ฅ ๐ก then the output will be ๐ฆ1 ๐ก = ๐๐๐ ๐ ๐ฅ ๐ก = ๐๐๐ ๐ + log(๐ฅ ๐ก ) ≠ ๐ ๐ฆ(๐ก). Therefore, the system is not homogenous • ๐ฆ ๐ =๐ฅ ๐ +5 • If we apply a new input ๐ฅ1 [๐] = ๐ ๐ฅ[๐] then the output will be ๐ฆ1 [๐] = ๐ฅ1 ๐ + 5 = ๐ ๐ฅ ๐ ๐ ๐ฆ ๐ = ๐ ๐ฅ ๐ + 5 = ๐ ๐ฅ ๐ + 5๐ ≠ ๐ฆ1 [๐]. Therefore, the system is not homogenous + 5. Additivity/Superposition • Let ๐ฅ1 (๐ก) and ๐ฅ2 (๐ก) be two different inputs to a system and let ๐ฆ1 (๐ก) and ๐ฆ2 ๐ก be their corresponding outputs. • A system is said to have additivity property if an input ๐ฅ1 ๐ก + ๐ฅ2 (๐ก) produces an output ๐ฆ1 ๐ก + ๐ฆ2 ๐ก . • In other words, if a system possesses additive property, the output(response) of the system corresponding to a sum of two inputs will be the sum of the individual outputs generated by these inputs. • Note that if the above is true for two inputs, then it can be generalised to n inputs • If the outputs corresponding to the inputs ๐ฅ1 (๐ก) , ๐ฅ2 ๐ก , …, ๐ฅ๐ (๐ก) are ๐ฆ1 (๐ก), ๐ฆ2 ๐ก , … ๐ฆ๐ ๐ก then the input ๐ฅ1 ๐ก + ๐ฅ2 ๐ก + โฏ + ๐ฅ๐ ๐ก produces an output ๐ฆ1 ๐ก + ๐ฆ2 ๐ก + โฏ + ๐ฆ๐ ๐ก . Additivity/Superposition • Similarly, we can define additive property for discrete time systems also. Let ๐ฅ1 [๐] and ๐ฅ2 [๐] be two different inputs to a system and let ๐ฆ1 [๐] and ๐ฆ2 [๐] be their corresponding outputs. • A system is said to have additivity property if an input ๐ฅ1 [๐] + ๐ฅ2 [๐] produces an output ๐ฆ1 [๐] + ๐ฆ2 [๐]. • If the outputs corresponding to the inputs ๐ฅ1 [๐] , ๐ฅ2 [๐], …, ๐ฅ๐ [๐] are ๐ฆ1 [๐], ๐ฆ2 [๐], … ๐ฆ๐ [๐] then the input ๐ฅ1 [๐] + ๐ฅ2 [๐] + โฏ + ๐ฅ๐ [๐] produces an output ๐ฆ1 [๐]+ ๐ฆ2 ๐ + โฏ + ๐ฆ๐ [๐]. Example • ๐ฆ ๐ก = 3๐ฅ ๐ก − 2 • Let ๐ฅ1 (๐ก) and ๐ฅ2 (๐ก) be two different inputs to this system then their corresponding outputs are given by. • ๐ฆ1 ๐ก = 3๐ฅ1 (๐ก − 2), ๐ฆ2 ๐ก = 3๐ฅ2 (๐ก − 2) • Now we apply an input ๐ ๐ก = ๐ฅ1 ๐ก + ๐ฅ2 (๐ก). Let the output be q(t). • ๐ ๐ก = 3๐ ๐ก − 2 = 3 ๐ฅ1 ๐ก − 2 + ๐ฅ2 ๐ก − 2 • Hence the system has additivity property = 3๐ฅ1 ๐ก − 2 + 3๐ฅ2 ๐ก − 2 = ๐ฆ1 ๐ก + ๐ฆ2 ๐ก Example • ๐ฆ ๐ก = ๐ฅ2 ๐ก • Let ๐ฅ1 (๐ก) and ๐ฅ2 (๐ก) be two different inputs to this system then their corresponding outputs are given by. • ๐ฆ1 ๐ก = ๐ฅ1 2 (๐ก), ๐ฆ2 ๐ก = ๐ฅ2 2 (๐ก) • Now we apply an input ๐ ๐ก = ๐ฅ1 ๐ก + ๐ฅ2 (๐ก). Let the output be q(t). • ๐ ๐ก = ๐2 ๐ก = ๐ฅ1 ๐ก + ๐ฅ2 ๐ก 2 = ๐ฅ1 2 ๐ก + ๐ฅ2 2 ๐ก + 2๐ฅ1 (๐ก) ๐ฅ2 (๐ก) ≠ ๐ฆ1 ๐ก + ๐ฆ2 ๐ก • Hence the system does not have additivity property Example • ๐ฆ ๐ =๐ฅ ๐ +2 • Let ๐ฅ1 [๐] and ๐ฅ2 [๐] be two different inputs to this system then their corresponding outputs are given by. • ๐ฆ1 ๐ = ๐ฅ1 ๐ + 2, ๐ฆ2 ๐ = ๐ฅ2 ๐ + 2 • Now we apply an input ๐[๐] = ๐ฅ1 [๐] + ๐ฅ2 ๐ . Let the output be q[n]. • ๐ ๐ = ๐ ๐ + 2 = ๐ฅ1 ๐ + ๐ฅ2 ๐ + 2 ≠ ๐ฆ1 [๐] + ๐ฆ2 [๐] • Hence the system does not have additivity property Practice Problems • Check if the following systems are additive or not ๐ฆ ๐ก = ๐ฅ ๐ก+3 ๐ฆ ๐ก = log(๐ฅ(๐ก)) ๐ฆ ๐ก = ๐ฅ(2๐ก) ๐ฆ ๐ = sin(๐ฅ ๐ ) ๐ฆ[๐] = ๐ฅ[2๐ + 3] ๐ฆ[๐] = ๐ ๐ฅ[๐] Practice Problems • Check if the following systems are additive or not ๐ฆ ๐ก = ๐ฅ ๐ก + 3 -yes ๐ฆ ๐ก = log(๐ฅ(๐ก))-no ๐ฆ ๐ก = ๐ฅ(2๐ก)-yes ๐ฆ ๐ = sin(๐ฅ ๐ )-no ๐ฆ[๐] = ๐ฅ[2๐ + 3]-yes ๐ฆ[๐] = ๐ ๐ฅ[๐] -no Stability of a system • A stable system is one in which small inputs do not lead to outputs/responses that diverge to infinity. • If the response of a system grows unbounded when the input is small and bounded, then the system is said to be unstable. • A Bounded Input Bounded Output (BIBO) stable system is one in which the output is always bounded to be less than a limit when the input is guaranteed to be less than a limit. • Mathematically if |๐ฅ(๐ก)| ≤ ๐ต๐ , then |๐ฆ(๐ก)| ≤ ๐ต๐ for all time instances for a Bounded Input Bounded Output (BIBO) stable system where ๐ต๐ and ๐ต๐ are constants. • If the above is not true for a system, we say that the system is BIBO unstable. • Similarly for discrete time systems if |๐ฅ[๐]| ≤ ๐ต๐ , produces an output |๐ฆ[๐]| ≤ ๐ต๐ for all time instances n, then it is a Bounded Input Bounded Output (BIBO) stable system. Example • ๐ฆ ๐ = σ๐๐=−∞ ๐ฅ ๐ • Assume that the input signal is bounded |๐ฅ[๐]| ≤ ๐ต๐ • |๐ฆ ๐ | = | σ๐๐=−∞ ๐ฅ ๐ | ≤ σ๐๐=−∞ ๐ฅ ๐ ≤ σ๐๐=−∞ ๐ต๐ = ∞ • |๐ฆ ๐ | becomes ∞ if we assume ๐ฅ ๐ = ๐ต๐ for all n. There fore the system is not BIBO stable. Example • ๐ฆ ๐ = 0.5 ๐ฅ ๐ + 2 ๐ฅ ๐ − 3 • Assume that the input signal is bounded ๐ฅ ๐ ≤ ๐ต๐ • ๐ฆ ๐ = 0.5 ๐ฅ ๐ + 2 ๐ฅ ๐ − 3 ≤ 0.5 ๐ฅ ๐ + 2 ๐ฅ ๐ − 3 ≤ 0.5๐ต๐ + 2๐ต๐ = 2.5๐ต๐ = ๐ต๐ • |๐ฆ ๐ | becomes 2.5๐ต๐ if we assume ๐ฅ ๐ = ๐ต๐ for all n. There fore the system is BIBO stable. Example • ๐ฆ(๐ก) = ๐ ๐ฅ(๐ก) • Assume that the input signal is bounded ๐ฅ(๐ก) ≤ ๐ต๐ • ๐ฆ(๐ก) = ๐ ๐ฅ(๐ก) = ๐ ๐ฅ(๐ก) ≤ ๐ ๐ต๐ = ๐ต๐ • |๐ฆ(๐ก)| becomes ๐ ๐ต๐ if we assume ๐ฅ(๐ก) = ๐ต๐ . There fore the system is BIBO stable. Example • ๐ฆ(๐ก) = ๐ก๐ฅ(๐ก) • Assume that the input signal is bounded ๐ฅ(๐ก) ≤ ๐ต๐ • ๐ฆ(๐ก) = ๐ก๐ฅ(๐ก) = ๐ก |๐ฅ(๐ก)| ≤ |๐ก|๐ต๐ • Output does not reach ∞ as long as t is finite. But if t->∞ the output also diverges to ∞. • Hence the system is not BIBO stable. Practice Problems • Check if the following systems are BIBO stable or not ๐ฆ ๐ก = ๐ฅ 2๐ก + 5 ๐ก ๐ฆ ๐ก = เถฑ ๐ฅ ๐ ๐๐ −∞ ๐ก2 ๐ฆ ๐ก = ๐ฅ(๐ก + 2) 2 ๐ฆ ๐ =3๐ฅ ๐ ๐ฆ ๐ = ๐ ๐ฅ[๐ − 4] ๐ฆ ๐ = ๐ฅ ๐ − ๐ฅ[๐ − 1] Practice Problems • Check if the following systems are BIBO stable or not ๐ฆ ๐ก = ๐ฅ 2๐ก + 5 -yes ๐ก ๐ฆ ๐ก = โซืฌโฌ−∞ ๐ฅ ๐ ๐๐-no ๐ฆ ๐ก = ๐ก2 2 ๐ฅ(๐ก + 2)-no ๐ฆ ๐ = 3 ๐ฅ ๐ -yes ๐ฆ ๐ = ๐ ๐ฅ[๐ − 4]-no ๐ฆ ๐ = ๐ฅ ๐ − ๐ฅ[๐ − 1]--yes