This article studies the joint optimization problem of computation offloading and resource allocation (JCORA) in mobile-edge computing (MEC). Deep reinforcement learning (DRL) is one of the ideal techniques for addressing the dynamic JCORA problem. However, it is still challenging to adapt traditional DRL methods for the problem since they usually lead to slow and unstable convergence in model training. To this end, we propose a temporal attentional deterministic policy gradient (TADPG) to tackle JCORA. Based on the deep deterministic policy gradient (DDPG), TADPG has two significant features. First, a temporal feature extraction network consisting of a 1-D convolution (Conv1D) residual block and an attentional long short-term memory (LSTM) network is designed, which is beneficial to high-quality state representation and function approximation. Second, a rank-based prioritized experience replay (rPER) method is devised to accelerate and stabilize the convergence of model training. Experimental results demonstrate that the decentralized TADPG-based mechanism can achieve more efficient JCORA performance than the centralized one, and the proposed TADPG outperforms a number of state-of-the-art DRL agents in terms of the task completion time and energy consumption.

Unmanned aerial vehicle (UAV) assisted emergency communication is an important technique for future B5G/6G scenario. The UAV is usually considered as a mobile relay to forward information from the macro base station (MBS) to the users in emergency area. In this paper, the MBS power allocation, the UAV service zone selection, and the user scheduling are jointly investigated to enhance the sum spectrum efficiency. We formulate the MBS power allocation and UAV service zone selection problem as an Markov Decision Process (MDP) in the delay ignored system (DIS) and propose a deep reinforcement learning (DRL) algorithm based on Q-learning and Convolutional Neural Networks (CNN). Then the proposed DRL-based scheme is extended in time delay system (TDS) to estimate the current optimal action with the outdated channel information. We also formulate the user scheduling as a 0-1 optimization problem and solve it by dividing into sub-problems. Simulation results demonstrate that the proposed DRL-based resource scheduling scheme can effectively improve the spectrum efficiency compared with the existing schemes.

Mobile edge computing (MEC) enhanced satellite based internet of things (SAT-IoT) is an important complement for terrestrial networks based IoT, especially for the remote and depopulated areas. For MEC enhanced SAT-IoT networks with multiple satellites and multiple satellite gateways, the coupled user association, offloading decision, computing and communication resource allocation should be jointly optimized to minimize the latency and energy cost. In this paper, the latency and energy optimization for MEC enhanced SAT-IoT networks are formulated as a dynamic mixed-integer programming problem, which is hard to obtain the optimal solutions. To tackle this problem, we decompose the complex problem into two sub-problems. The first one is computing and communication resource allocation with fixed user association and offloading decision, and the second one is joint user association and offloading with optimal resource allocation. For the sub-problem of resource allocation, the optimal solution is proven to be obtained based on Lagrange multiplier method. And then, the second sub-problem is further formulated as a Markov decision process (MDP), and a joint user association and offloading decision with optimal resource allocation (JUAOD-ORA) is proposed based on deep reinforcement learning (DRL). Simulation results show that the proposed approach can achieve better long-term reward in terms of latency and energy cost.

This paper proposes a telecom big data based user offloading self-optimisation (TBDUOS) scheme. Its aim is to assist telecom operators to effectively balancing the load distribution with achieving good service performance and customer management in heterogeneous relay cellular systems. To achieve these objectives, in the cell-level offloaded traffic analysis stage, the optimal offloaded traffic is calculated to minimise the total blocking probability. In the user-level offloading stage, the user portrait is drawn and the K-MEANS algorithm is employed to manage the users clustering in the heavily loaded cell, and finally shifting users to assistant cells. Simulation results show the TBDUOS scheme can effectively reduce the handover failure and call dropping of specific users, especially voice/stream users, high consumption users, high level users. The TBDUOS scheme can also reduce the blocking probability.

Existing localization methods that intensively leverage the environment-specific received signal strength (RSS) or channel state information (CSI) of wireless signals are rather accurate in certain environments. However, these methods, whether based on pure statistical signal processing or data-driven approaches, often struggle to generalize to new environments, which results in considerable time and effort being wasted. To address this challenge, we propose MetaLoc, which is the first fingerprinting-based localization framework that leverages the Model-Agnostic Meta-Learning (MAML). Specifically, built on a deep neural network with strong representation capabilities, MetaLoc is trained on historical data sourced from well-calibrated environments, employing a two-loop optimization mechanism to obtain the meta-parameters. These meta-parameters act as the initialization for quick adaptation in new environments, reducing the need for much human effort. The framework introduces two paradigms for the optimization of meta-parameters: a centralized paradigm that simplifies the process by sharing data from all historical environments, and a distributed paradigm that maintains data privacy by training meta-parameters for each specific environment separately. Furthermore, the advanced distributed paradigm modifies the vanilla MAML loss function to ensure that the reduction of loss occurs in a consistent direction across various training domains, thus facilitating faster convergence during training. Our experiments on both synthetic and real datasets demonstrate that MetaLoc outperforms baseline methods in terms of localization accuracy, robustness, and cost-effectiveness. The code and datasets used in this study are publicly available.

In recent years, the LTE/5G mobile networks operation and telecom services lead to the explosion of telecom big data. Since telecom big data stores valuable information, this paper researches on telecom big data and further designs a telecom big data platform, in order to effectively utilizing telecom big data resources for the application. The designed platform consists of five layers, including telecom data sources layer, telecom data integration layer, telecom data storage and calculation layer, operation and management layer, telecom data application layer. Based on this telecom big data platform, we design both the cell judgment system and the telecom customer analysis system, which are further applied in the local telecom operators.