A variety of waves are used in a wide range of technologies. For example, low frequency waves are used in geophysical exploration, microwaves are used in radar and wireless technologies and light waves are used in optical technologies. Prior to hardware developments of such highly technical devices it is important to understand how waves propagate through a certain space or in such devices.
Numerical simulations of the electromagnetic wave propagation give us intuitive information. The Finite Difference Time Domain (FDTD) method is the most common method of modelling the propagation of waves due to its simplicity and versatility. There are many companies which sell the simulation software based on the FDTD method. However, modelling the objects whose size is less than wavelength/10000 in an electrically large FDTD space is a challenge to any researchers in companies and research institutes as the requirement of memory and time for computation can easily go over several TB and several weeks.
Together with the world leading researchers such as Jean-Pierre Berenger in France, Salvador Garcia in Spain, Mohamed Latrach in France, Costas Sarris in Canada, Mehmet Yavuz in the US, Ryutaro Himeno in Japan, Seiji Fujino in Japan, and companies such as REMCOM in the US and 2COMU in the US, we work for the improvement of the computational efficiency of the FDTD method mainly for the human body and graphene modelling. Very high absorbent and computationally efficient Huygens absorbing boundary condition (patent in the US), subcell and subgridding methods, filtered FDTD, implicit schemes such as Crank-Nicolson, alternating direction implicit, locally one dimensional, are the ones we deals with as the algorithmic approaches.
As for the hardware approach, we are developing the most efficient way to parallelise our FDTD code for the shared memory architectures (OpenMP), the distributed memory architectures(MPI), and Graphics Processing Unit(GPU) with the register-level parallelisation(SSE, AVX).
Using the FDTD method, we also study the time reversal algorithm to enable it to be usable for the frequency dependent materials such as human tissues for cancer localisation/ablation for example. One of the ways to treat the Parkinson's disease is the stimulation of the SubThalamic Nucleus (STN) by inserting an electrode invasively. In order to produce non-invasive deep brain stimulation, the time reversal algorithm needs the waveforms observed at the skull when STN is stimulated invasively.Thus we solved the forward problem.
Image ppm-plane-x-00000.jpg shows the excitation point which is in the middle of STN, ppm-plane-z-00001.jpg is another cross section of the excitation point and ppm-plane-z-00028.jpg is 28 mm below ppm-plane-z-00001.jpg .
For the Debye media parameter setting for each human tissue, we fitted the data measured by the US Air Force to the one-pole Debye model and presented here.
Figure 15 shows how wave propagates from STN. The signal comes out of eyes first and second from the left ear (due to the excitation of the left STN) and reach the observation-plane, 50% of the cancer patients have unresectable cancers.
They have to take chemotherapy. However, half of those patients do not respond to the chemotherapy. Hyperthemia treatment improves the efficacy of the chemotherapy. Chemo-middle.avi observes the field distribution at the center of the human body.