Energy-Efficient Secure Video Streaming in UAV-Enabled Wireless Networks A Safe-DQN Approach
ABSTARCT :
Unmanned aerial vehicles (UAVs) are anticipated to be integrated into the next generation wireless networks as new aerial mobile nodes, which can provide various live streaming applications such as surveillance, reconnaissance, etc. For such applications, due to the dynamic characteristics of traffic and wireless channels, how to guarantee the quality of service (QoS) is a challenging task. In this paper, with recent advances in scalable video coding (SVC), we study energy-efficient secure video streaming in rotary-wing UAV-enabled wireless networks. By jointly optimizing video levels selection, power allocation, and the UAV’s trajectory, we intend to maximize the long-term energy efficiency that is defined as the ratio of video quality to power consumption. Meanwhile, secrecy timeout probability is considered as a constraint cost to guarantee time delays requirements in a long run perspective. Our problem is modeled as a constrained Markov decision process (CMDP) and solved by safe deep Q-learning network (safe-DQN), where a safe policies set induced by constructing a Lyapunov function is dynamically adjusted to satisfy the constraint conditions of the CMDP. Extensive simulation results with different system parameters show the effectiveness of the proposed algorithm compared with other existing reinforcement learning algorithms.
EXISTING SYSTEM :
? Special emphasis was placed on providing constructive criticism to some existing works and on exploring open issues to the reader.
? We will also highlight the limits of the existing works and outline some potential future applications of ML for UAVs networks.
? With the vast amount of published work linking machine learning to UAV wireless networks, several tutorials and surveys have attempted to summarize the existing literature.
? The first step consists in detecting potential UAV existing by analyzing the frequency and then check whether the sound exceeds a predefined threshold for drones.
? To the work mentioned above, there exist other tutorials and surveys that are oriented towards the application of ML tools in wireless communication networks.
DISADVANTAGE :
? This framework breaks the complicated optimization problem into two sub-problems and solves the sub-problems in an iterative manner.
? Game theory provides tools to solve multi-agent decision problems and to analyze the interactions among various agents in a communication network.
? In particular, the authors proposed Mean-Field Games (MFG) to solve problems in massive UAVs networks.
? For the A2G channel modeling problem as an example, there exists no specific modeling method for channel measurements in urban areas and rural areas under various weather conditions.
PROPOSED SYSTEM :
• A shallow artificial neural network is proposed to analyze the effect of several natural phenomena on the signal such as : diffraction, reflection, and scattering.
• While classical RL proposed an efficient solution for many types of discrete decision problems, more realistic solutions could be provided using DRL which proven its efficiency by reaching super human level control.
• It is worth to mention that DQN was proposed as an improvement to Q-learning which uses a discrete state and action space in order to build the Q-table.
• The proposed modelfree RL-assisted framework enables dynamic tracking of users’ movement by adjusting the UAV location accordingly.
ADVANTAGE :
? The height and elevation angle of a UAV impact its coverage performance and link reliability over a service area.
? UAVs can boost the performance of existing ground wireless networks in terms of coverage, capacity, delay and overall quality of service.
? With the development of high performance computing hardware and the availability of large data sets, Machine Learning (ML) techniques have recently been applied to many fields due to their ability of “learning” from interacting with the environment.
? This method reduces the time complexity of solving such problems while maintaining as good performance as the classical MILP methods.
? With this mechanism, on one hand, a loss of one UAV’s data will not greatly affect the whole system performance.
|