As the demand on time-sensitive contents increases, data freshness recently becomes an important performance metric in the cache-enabled networks. Therefore, in this paper, we design the joint cache replacement and content update algorithm in the cloud-edge networks considering the data freshness at both the cloud server and the edge server. We define a fresh content acquisition with cache hit (FACH) ratio as a performance metric, which shows the portion of users obtaining the requesting content from the edge server while satisfying the freshness constraint. To maximize the FACH ratio, we propose the reinforcement learning (RL)-based algorithm, named theintelligent Cache replacement and content Update(i-CU)algorithm. In the proposed algorithm, we newly suggest the score-based action decision to reduce the action space while guaranteeing the constraints of the problem. In the simulation results, we develop and evaluate thei-CUalgorithm for various datasets, which verifies that thei-CUalgorithm can achieve the higher FACH ratio compared to the existing baselines under the various network parameters.

Since the permissioned blockchain technology has been proposed to ensure data integrity in distributed systems, the low throughput and high latency have been recognized as major issues. In some applications, the data, available later than allowed time, can be useless, so the effective throughput is newly considered, defined as the average number of transactions per second, committed within the required latencies. For maximizing the effective throughput, we propose a novel intelligent transaction generation control (i-TGC) method to determine the transaction generation for each client. To improve performance in the dynamic environment of blockchain services based on real-time information, we employ the reinforcement learning (RL) for the i-TGC algorithm. Our experiment results show the i-TGC outperforms the probabilistic transaction generation control (p-TGC), which generates transactions randomly with the optimal probability that maximizes the effective throughput. We also verify the performance of the i-TGC for various environments with different block sizes, block generation timeout, traffic patterns, and the number of clients. The i-TGC can be a way to accelerate the usage of the permissioned blockchain for latency-sensitive services.

As the demand on artificial intelligence (AI)-based applications increases, the freshness of sensed data becomes crucial in the wireless sensor networks. Since those applications require a large amount of computation for processing the sensed data, it is essential to offload the computation load to the edge computing (EC) server. In this paper, we propose the sensing and computing decision (SCD) algorithms for data freshness in the EC-enabled wireless sensor networks. We define the η-coverage probability to show the probability of maintaining fresh data for more than η ratio of the network, where the spatial-temporal correlation of information is considered. We then propose the probability-based SCD for the single pre-charged sensor case with providing the optimal point after deriving the η-coverage probability. We also propose the reinforcement learning (RL)-based SCD by training the SCD policy of sensors for both the single pre-charged and multiple energy harvesting (EH) sensor cases, to make a real-time decision based on its observation. Our simulation results verify the performance of the proposed algorithms under various environment settings, and show that the RL-based SCD algorithm achieves higher performance compared to baseline algorithms for both the single pre-charged sensor and multiple EH sensor cases.

In this paper, we propose a joint sensing and computing decision algorithm for data freshness in edge computing (EC)-enabled wireless sensor networks. By introducing the data freshness at the presented networks, we define the$\eta$-coverage probability to show the probability of maintaining fresh data for more than$\eta$ratio of the network, where the spatial-temporal correlation of information is considered. To maximize the$\eta$-coverage probability in the networks with limited energy, we propose the reinforcement learning (RL)-based decision algorithm by training the policy of sensors. Our simulation results verify the performance of the proposed algorithm for different number of sensors and the computing energy. From the results, we show the proposed algorithm achieves higher$\eta$-coverage probability compared to the baseline algorithms.