Hardware-Efficient Root Value Estimation for Zadoff-Chu Sequences of Known Length

Sammanfattning

In wireless communication, Zadoff-Chu (ZC) sequences are often used for synchronisation between transmitters and receivers due to their advantageous mathematical properties. They are extensively used in LTE and 5G systems, as well as for UAV communications and various other applications. Blind detection of ZC sequences, meaning that their structural parameters are unknown, has applications for security and privacy. However, blind ZC sequence detection is computationally expensive, especially for longer sequences, restricting deployment to centralised processing platforms. This thesis investigates hardware efficient algorithms for the detection of ZC sequences for deployment on resource-constrained edge devices. We simulated and evaluated three algorithms: time-based cross-correlation, fast cross-correlation, and phase difference estimation. The results demonstrated that both cross-correlation algorithms achieved acceptable performance down to a Signal-to-Noise Ratio (SNR) of -14 decibels. However, the time-based cross-correlation consumed 10 times the energy, rendering it impractical. In contrast, the phase difference estimation algorithm consumed 10 times less energy than the fast cross-correlation algorithm, but the acceptable performance was limited to SNRs above -6 dB. To further reduce energy consumption, these algorithms were also modified. This yielded significant power savings proportional to the modification degree; however, performance decreased accordingly. Consequently, these trade-offs demonstrate that the optimal algorithm and modification configuration depend on the intended SNR environment, providing a dynamic low-power solution.