Effective Energy Efficiency and Statistical QoS Provisioning Under Markovian Arrivals and Finite Blocklength Regime

In this article, we evaluate the effective energy efficiency (EEE) and propose delay-outage aware resource allocation strategies for energy-limited Internet of Things (IoT) devices under the finite blocklength (FBL) regime. The EEE is a cross-layer model, measured by the ratio of effective capacity to the total consumed power. To maximize the EEE, there is a need to optimize transmission parameters, such as transmission power and rate efficiently. Whereas it is quite complex to study the impact of transmission power, or rate alone, the complexity is aggravated by the simultaneous consideration of both variables. Hence, we formulate power allocation (PA) and rate allocation (RA) optimization problems individually and jointly to maximize EEE. Furthermore, we investigate the performance of the EEE under constant and random arrivals, where statistical QoS constraints are imposed on buffer overflow probability. Using effective bandwidth and effective capacity theories, we determine the arrival rate and the required service rate that satisfy the QoS constraints. After that, we compare the performance of different iterative algorithms, such as Dinkelbach's and cross entropy, which guarantee the convergence for the optimal solution. By numerical analysis, the influence of source characteristics, fixed transmission rate, error probability, coding blocklength, and QoS constraints on the throughput are identified. Our analysis reveals that the joint PA and RA is the optimal resources allocation strategy for maximizing the EEE in the presence of constant and random data arrivals. Finally, the results illustrate that modified Dinkelbach's algorithm has high performance and low complexity compared to others.