Cross-Layer Design Toward Ultra-Reliable And Low-Latency Communications

With the rise of many time-sensitive applications,ultra-reliable and low latency communications(URLLC),as one main scenario of the fifth generation wireless communications(5G),are extensively studied in the recent years.Existing works mainly focus on employing short-frame structure to satisfy the ultra-reliable and low-latency requirement over slow fading channels.The short-frame is with respect to the finite blocklength.The formula of channel capacity in the finite blocklength implies that the achievable rate decreases as blocklength decreases.Thus,the queuing delay increases as the frame-length decreases.The short frame-length also leads to small transmission delay.There exists a tradeoff between queuing delay and transmission delay.The fixed short-frame structure cannot well balance the queuing delay and transmission delay,which results in the high end-to-end delay.Also,how to satisfy the ultra-reliable and low-latency requirement over fast fading channels is an open issue.To deal with the above-mentioned issues,the adaptive frame-length scheme and transmission power adaptive scheme over fast fading channels are proposed in this thesis.The detailed works are as follows:1.The Adaptive Frame-Length Scheme.To achieve the optimal balance between queuing delay and transmission delay,the average end-to-end delay minimization problems are studied.For single user case,a variable transmission time interval based queuing model is proposed to adapt to the time-varying arrival of packets.With this queuing model,the formulation of the average end-to-end delay minimization problem for single user is derived.It is nonconvex and difficult to solve.To tackle this problem,a flexible proximal alternating direction method of multipliers is employed.Using it the optimal adaptive frame-length scheme is proposed.Numerical results show that the proposed adaptive frame-length scheme can significantly reduce the average end-to-end delay compared with the traditional fixed frame-length scheme.The average end-to-end delay minimization problem for multi-user case is also studied,where the subchannel allocation is considered.Due to the discreteness of subchannel allocation,the corresponding problem is highly complex.To tackle this problem,a multiple deep Q-learning networks based adaptive subchannel allocation and frame-length design scheme is proposed.Numerical results show that the proposed scheme can achieve the better balance between queuing delay and transmission delay than the fixed frame-length and adaptive subchannel allocation schemes in long-term evaluation and 5G new radio.2.The Adaptive Transmit Power Scheme for Fast Fading Channels.In fast fading circumstances,the instantaneous channel state information(CSI)feedback results in the low latency requirements in URLLC hard to be satisfied.In this paper,the transmit power minimization problem where only statistical CSI is known at the transmitter is studied for URLLC.Solving this problem,the optimal adaptive transmit power scheme is derived.For badly fading channels,the required power is still large.To further reduce the minimum required power,the multiple antennas technique is employed to improve the channel quality.Whereas,the outage probability of multiple input multiple output channel is very complex.A pair of upper-bound and lower-bound for the outage probability is proposed.Substituting the boundaries into the transmit power minimization problem,the upper-bound and low-bound of optimal transmit power is derived.Numerical results show that the derived upper-bound is very close to the optimal transmit power.