One classical solution is based on a PLL - which, in this case, can be regarded as a very small "moving" bandpass (the center frequency automatically adapts to the receive frequency).

Thank you for the answer. This PPL method, is it done in the analog or digital domain? I will look for some tutorial about it.

Is it also a good idea to send a small chirp before each transmission to make it easier to search with a matched filter?

PLL is usually implemented d in analog domain but digital locked loop also exist that do the same in discrete time domain. however the most useful techniques for your case would be frequency recovery technique in addition to clock and phase recovery. the book written by Mengali is great reference in synchronization technique.

In a vehicular networking communication scenario, an effective way to suppress the doppler effect is to use low frequency channels and tighten the channel width. This has been proven to be effective through various studies in the literature. I hope this helps.

I am not sure that idea for work UAV system, because of the size of UAV, they probably have limitation on size, bandwidth and carrier frequency of their device. tightening the bandwidth means decreasing the capacity and bandwidth efficiency of UAV communication system which is against the power saving policy of a UAV. so i guess implementing a state of the art synchronization would be the best choice!

Hello, you can always include a training sequence (preamble, mid amble or post amble ) like in GSM . On the receiver side, you will do Autocovariance of the received part together with the locally generated training sequence. In that you will know exactly how many time slots you should consider to delay in time domain.

I agree with Behrad, the way to go is a good synchronisations solution most likely based around a PLL setup with the right parameters to provide the lowest noise bandwidth while allowing to track the frequency variations expected by the relatively high speed of the UAV. I have found the Mengali reference book to be the best I've seen on the topics you are interested in. I have recently designed a UAV communications link for the SARape European Defence Agency contract (http://www.project-sarape.eu/index.php?option=com_content&view=article&id=9&Itemid=15) and have used a combination of UW detectors, Muller & Mueller symbol timing recovery, and a second order PLL for Frequency and Phase Estimation. Of course the algorithms of choice will depend on the waveform and air interface standard used, which could be continuous or busty, which will make a significant difference to the algorithm choices. The tracking capability of the algorithms is also very significant in wide bandwidth and fast manoeuvring vehicles. Keep in mind that multipath fades can affect your system significantly, so make sure this is handled by the algorithms (or waveform) you select for your purpose.

Christopher, did you implement your modem on DSP or FPGA platform ? I agree all your comments however M&M is not best choice for UAV, because of the low convergence with respect to other state of the art clock recovery like near maximum likelihood and early late gate technique. though these techniques need over sampling and interpolation filters, their steady state and convergence speed outperf M&M. actually using over sampling at the ADC level or baseband level ( FPGA level) is recently became very feasible.

Was second order PLL sufficiently for your highly dynamic UAV? did you have any adaptive filter for your frequency selective fast fading or you have chosen to use differential modulation techniques for UW ?

Behrad, I agree totally with your comment on the M&M, it's not he fastest converging, but as always with these design trade-offs it all depends on what you place emphasis on, speed, simplicity, performance etc. In my case, the design was implemented on FPGA at had to process 100MHz wide carrier within a small FPGA. Oversampling and interpolation was not a choice so better timing recovery methods where not selected. the second order PLL was certainly sufficient in this instance given that the UAV although having high speed did not impose fast manoeuvring which would cause issues in tracking. Of course the issue of loop bandwidth noise was set to tolerable levels to allow for the expected Doppler effects vs. frequency/phase tracking performance. I would have liked to have used an adaptive filter but again power and complexity limitations demanded a simpler solution, so the fading was handled mostly by an OFDM waveform combined with AGC management, distributed UW's and of course data stream coding. It was initially thought to use a simple air interface loosely based on the DVB standards, and use a modulation scheme combined with an adaptive filter. Finally however, the OFDM alternative was preferred in terms of complexity of the design vs. performance, so deviating from the baseline DVB-S choice. DVB-RCS, DVB-S2, UMTS and others where also considered, although the simplicity of the design needed forced for a more custom solution.

Overall I was surprised with how well the system worked given the simplicity of the choices I have made. The design was small, fitting all the Rx logic into a Xilinx FPGA using only 8K LUTS, 12 block rams, and 40 DSP48's, and could handle a carrier BW of up to 150MHz. To get it this small I used DSP sharing combined with clever RAM management and extensive fixed point precision management techniques in all the arithmetic. The ADC used is 10bit, sampling at 400MHz.

I would be interested to hear your views and what other choices you would have made given the small size design targets.

your solution for UAV is quite state of the art because of you knowledge about all those of techniques and also FPGA design and synthesis. actually your high data rate and broadband system surprised me because i am already aware of the hardship to tachkle 400Mhz clock in FPGA domain. since you are aware of the DSP module and gates of your implemented system on FPGA you were probably involved with the FPGA implementation. i appreciate your detail description of the system you have implemented. i am a little confused that you have finnaly chosed UW or OFDM system because these two have different goals and challenges.

did you face microphonic noise in your system ? though you answer was very detailed but your choice of modulations and coding is not put forward. ( OFDMA and UW together with QAM , differential modulation, PSK , DPSK , TCM or high end Adaptive modulation and coding scheme)

at the moment I am working on super high vision video transmission over highly dynamic channel like UAV but with a rich hardware platform (four veritex V board). our data bandwidth is over 1 GHz that normal ADC can not cover this rate and micro-phonic noise\clock drift is one of the main challenges in this project because for performance efficiency we aim to use coherent modulation in conjugation with turbo products codes. generally mechanical shock that frequently appear in UAV system result in micro-phonic noise. one of the main directives of our project goals is not to have any bit error due to environmental problems of UAV .

we have chosen not to handle shocking clock drift by dangling the crystal and only work on this problem by baseband design and process.