On Fundamental Performance Limits of Delay-Sensitive Wireless Communications

Mobile traffic is expected to grow at an annual compound rate of 57% from 2014 to 2019, while among the data types that account for this growth mobile video has the highest growth rate. Since a significant portion of mobile video traffic is delay-sensitive, delay-sensitive traffic will play a critical role in future wireless communications. Consequently, future mobile wireless systems will face the dual challenge of supporting large traffic volume while providing reliable service for various kinds of delay-sensitive applications (e.g. real-time conversational video, voice-over-IP (VoIP), and online gaming). Past work on delay-sensitive wireless communications has generally overlooked physical-layer considerations such as modulation and coding scheme (MCS), probability of decoding error, and code blocklength (or coding delay) by employing oversimplified models for the physical-layer. In this dissertation we aim to bridge information theory, communication theory and queueing theory by jointly considering the queueing delay violation probability and the probability of decoding error to identify fundamental trade-offs among wireless system parameters such as MCS, code blocklength, user perceived quality of service, channel fading speed, and average signal-to-noise ratio (SNR). Throughout this dissertation we focus on wireless communication systems where the channel state information (CSI) is available only at the receiver. We model the underlying wireless channel by a finite-state Markov chain (FSMC) where state transitions happen at each transport block (TB) transmission (i.e. TB-based FSMC). First, we focus on communication schemes without feedback and derive the dispersion of the TB-based FSMC model of the Rayleigh fading channel. The TB-based FSMC dispersion is used to characterize the maximum achievable throughput under probability of decoding error and coding delay constraints for a given modulation scheme. Second, we focus on communication schemes with one bit decision-feedback (e.g. acknowledge (ACK)). We introduce a communication scheme, namely early decoding, where the receiver determines the decoding time based on the available CSI. We characterize the maximum achievable throughput of the early decoding scheme under probability of decoding error and coding delay constraints for a given modulation scheme. Then, we derive the dispersion of parallel additive white Gaussian noise (AWGN) channels with finite discrete input alphabets (e.g. pulse amplitude modulation (PAM)). The dispersion of parallel AWGN channels is used to track the operation of incremental redundancy type hybrid automatic repeat request (IR-HARQ) over the Rayleigh fading channel through the HARQ Markov model (HARQ-MM), introduced here. We use the HARQ-MM to characterize the maximum achievable (average) throughput of IR-HARQ under probability of decoding error and coding delay constraints for a given MCS. Third, we focus on a queueing system where data packets arrive at the transmitter, wait in the queue, and are transmitted over the Rayleigh fading channel with IR-HARQ. We invoke a two-dimensional discrete-time Markov process and develop a recursive algorithm to characterize the maximum achievable (average) system throughput for a given MCS under queueing delay violation probability, and probability of decoding error constraints