Design Of Active Noise Cancellation System For Headphones Based On FXLMS Algorithm

With the continuous development of society,the noise problem in the city is also increasing.There are many noises in urban life: the running sound of machines in factories,the engine sound and horn sound of motor vehicles on the road,the noise of music played on the street,etc.These noises will not only induce many bad emotions,such as anger,irritability,anxiety,etc.,but also cause insomnia in the listener,so that it has a very negative impact on work and study.Therefore,in order to solve the noise problem,noise control technology has been paid attention and developed rapidly.Noise control technology can be divided into two types: passive noise control technology and active noise control technology,both of which have noise reduction products developed by their noise reduction principles in the market.Passive noise reduction technology first appeared,because it not only has low design cost and simple principle,but also has a large noise reduction space.However,although passive noise reduction technology has a good control effect,its noise reduction frequency is only suitable for high noise.The disadvantage is that it has poor noise reduction effect for non-high frequency noise.Due to the limitations of passive noise reduction technology,the research and exploration of active noise reduction technology has begun to receive attention.Active noise reduction technology can not only overcome the noise reduction drawbacks of passive noise reduction,but also show better noise reduction performance.Active noise reduction technology can be understood as using the principle of superposition of sound waves to generate sound waves to cancel the noise generated in the space.The sound waves have the same amplitude and opposite phase as the noise.Active noise reduction can also be divided into feedforward noise reduction and feedback noise reduction.The structure of the two is exactly the same,but the system input is different.The former input is primary noise and the latter is residual noise.The feedforward noise reduction system is very stable but easily affected by the external environment,while the feedback noise reduction is not stable enough not to be disturbed by the external environment,and cannot achieve a large noise reduction bandwidth.Composite noise reduction can be seen as a combination of feedforward and feedback,and it has the advantages of both.The FXLMS（filtered-x least mean square）algorithm that controls the adaptive filter to update the weight in real time is the core of the active noise reduction system,and the performance of this algorithm will directly affect the noise reduction performance of the active noise reduction system.The LMS（Least Mean Square）algorithm has the advantages of being easy to implement,stable and good performance,and wide application range.The principle of the LMS algorithm can be understood as taking the weights updated by the weights as input,and then through the operation of the algorithm,and finally making the output error signal reach the minimum value of the operation sequence,and then re-update the next group according to the error value.Weights,and so on and so forth to minimize the error value.In the practical application of the design,the residual noise is collected by the error microphone as the system error input signal of the LMS algorithm,and the weight coefficient of the filter is modified in real time through the algorithm,and finally the purpose of minimizing the residual noise is realized.However,in the face of practical In the application scenario,the sound signal emitted by the system speaker is transmitted to the secondary channel generated by the error microphone,which will have a great impact on the performance of the algorithm,and will cause system divergence in severe cases.The FXLMS algorithm is an adaptive algorithm that takes the secondary channel into account.The method of this algorithm is to use an FIR（Finite Impulse Response）filter to simulate the secondary channel,which strengthens the system when facing the secondary channel.noise reduction effect.This paper mainly completes the following work:1.Through the study of the FXLMS algorithm,combined with the functions to be realized by the algorithm,three sub-modules have been written in matlab: Delay delay module,Filter filter module and LMS weight update module.Four operations are accomplished by combining these three sub-modules in the top-level module: filtering,secondary channel simulation,computing gradients,and weight update.According to these four operations,draw a block diagram of the algorithm structure to realize the functions that need to be completed.Finally,the work is completed through software simulation:（1）Through the collected noise input,the software is used to build a simulation model,and the noise reduction performance of the active noise reduction completed by the top-level module is verified.（2）Simulate the primary channel in matlab.（3）Build the headphone model through comsol,and finally obtain the transfer function of the actual sound in the ear canal of the headphone,and complete the offline modeling of the secondary channel in matlab according to this transfer function.2.Carry out hardware design according to the verified matlab model,use the sub-modules already defined in the matlab model,use the hardware language to complete the definition of the corresponding hardware sub-modules on modelsim,and instantiate the hardware sub-module design in the top-level module.Design an active noise reduction chip for headphones.The chip realizes the composite noise reduction of the FXLMS algorithm based on the offline modeling of the secondary channel,and innovatively designs the resource multiplexing algorithm.Feedforward noise reduction and feedback noise reduction are completed successively in the cycle,so that the integrated chip has the advantages of low power consumption and low area while realizing composite noise reduction.After simulation verification,the designed ANC chip has a certain robustness to various noises,and has a large noise reduction depth and noise reduction bandwidth.3.After the hardware design of active noise reduction is completed,the hardware model verification is first completed on the modelsim platform,and then the simulation waveform of the software is obtained by uploading the digital output results obtained from the hardware verification to matlab to verify the design.correctness.Upload the hardware code to the synthesis tool to generate the gate-level synthesis netlist and synthesis report of the chip.The comprehensive implementation designed in this paper is based on Global Foundry’s 0.18 um process,running at 48 MHZ clock,the chip power consumption is 19.0782 m W,the area is 667039,and the setup and hold time meet the requirements.After the design is completed,it undergoes multiple rounds of simulation testing and back-end synthesis.