Research On Age Of Information Based Power Allocation Policy For Communication Systems

Recently,with the rapid development of wireless mobile communication,the era of the Internet of Everything is coming.In this context,the demand for the timeliness of information in communications is becoming increasingly stringent.How to properly capture and describe the freshness of information in real-time communication systems has attracted great attention from scholars.Therefore,age of information(Ao I)has emerged as a new performance metric for measuring the freshness or timeliness of information and is rapidly becoming one of the research hotspots.This dissertation focuses on the optimization in terms of both peak age of information(PAo I)violation probability and average age of information.The main research is summarized as follows.· For the optimization problem of PAo I violation probability,a single-source wireless communication system coupled with the control application is analyzed.Based on the control-communication co-design approach,a Markov failure model is constructed,and mean time to failure(MTTF)is introduced as a performance metric for system reliability evaluation.Meanwhile,the optimization problem of PAo I violation probability is modeled as a constrained Markov decision process(CMDP)problem,in which different levels of transmit power can be selected based on the received feedback signal.For any given sampling rate,an optimal power allocation policy as well as a suboptimal threshold-based policy are proposed.Finally,numerical results show that the proposed policy can maintain the reliability of the system for hundreds of years and have a significant improvement in performance compared to the benchmark policy.· For the optimization problem of the weighted sum average Ao I,a multi-source wireless uplink communication system over block-fading multiple access channels is analyzed.Under the limit of maximum retransmission rounds,multiple independent sources send update packets to a common destination node.Different multiple access schemes,i.e.,orthogonal multiple access(OMA)and non-orthogonal multiple access(NOMA),are considered.The CMDP problems under two scenarios,i.e.,minimizing the weighted average Ao I under the average power constraint,are also analyzed.When the channel distribution information(CDI)is known at the transmitter side,with the help of the Lagrangian method,a CMDP problem can be converted into an equivalent unconstrained MDP problem.The optimal power allocation policy is derived by the corresponding offline value iteration algorithm.Through numerical results,it is verified that the proposed optimal policy reduces the weighted sum average Ao I significantly compared to the fixed power policy,and it is demonstrated that NOMA is more suitable for transmitting large update packets due to the higher spectral efficiency.· For the optimization problem of the weighted average Ao I in an unknown environment,the traditional offline value iteration algorithms are no longer feasible since CDI or error transmission probabilities are not available as valid pre-accessible information.Therefore,an online reinforcement learning method is proposed here.To overcome the problems of non-convergence or the curse of dimensionality in traditional Q-learning algorithms,a combination of Q-learning and ε-greedy exploration algorithm is designed.Numerical simulation results show that the proposed algorithm achieves an age performance close to the optimal policy and saves a lot of time by normalization.