Joint Power and Pilot Length Allocation for Ultra-Reliable and High-Throughput Transmission

Emerging services in Tactile Internet are expected to bring immersive experience for customers in watching live sports, action films, and advertisements, which demand ultra-reliable and high-throughput transmission. Since the power budget and the block length are limited for communications, allocating the transmit power and the pilot length becomes critical for efficient data transmission from the base station (BS) to the devices. In this work, taking the coding theory of finite blocklength and the resource allocation over channel estimation and data transmission into account, both data error probability and throughput are analyzed and optimized in terms of the pilot power and pilot length, respectively. It is revealed that both the data error probability and the throughput are quasiconcave and quasiconvex, respectively, with respect to ( $w.r.t.$ ) the pilot length and the power of pilot signal from a block-level perspective. Regarding a frame consisting of multiple blocks, we propose the truncated channel inversion scheme to allocate the data transmission power over multiple blocks. Given the length of pilots, the average data error probability and the average throughput for a frame are proved to be quasiconcave and quasiconvex, respectively, $w.r.t.$ the desired power level of the received data signal. We propose a sub-optimal algorithm for frame-wise throughput maximization to derive the length of the pilots and the desired power level of the received data signal. Numerical results show that the proposed scheme outperforms the fixed transmission power scheme in terms of the average data error probability for a frame by reducing an order of magnitude under the same system power constraints.