在未来自动驾驶系统中，车辆有效和安全的控制依赖于车速、路况等状态信息。这些随时间不断变化的状态信息由部署在车辆和环境中的传感器节点观测和采集。状态信息的时变性要求传感器源源不断进行状态信息采集，通过通信网络直接发送或由基站转发给用户。然而，网络中的随机传输时延、通信系统有限的带宽和传输速率、以及车辆移动所导致的信道时变性为采集和推送时效性强的状态信息带来了挑战。另一方面，好的状态信息时效性与传统通信质量要求，例如高吞吐量与低传输时延有本质不同。因此，我们需要重新设计传输和网络调度算法，优化网络中用户和决策中心的状态信息时效性，从而保证系统的有效和安全性。本文针对车联网中，基站通过传感器采集、用户从基站主动索取和被动推送状态信息的场景，研究优化状态信息时效性的信息采集和传输调度策略。论文依据状态信息更新特点，选用信息年龄和同步年龄度量信息时效性。主要贡献如下：1. 点对点链路信息年龄最小化的自适应状态信息采样。考虑状态信息传输存在随机时延，以最小化接收端平均信息年龄为目标，针对传输时延统计信息未知的挑战，设计了自适应在线采样策略。理论分析证明，随着时间和采样数目趋于无穷，算法得到的平均信息年龄趋于最优。2. 多传感器网络信息年龄最小化的采样、功率控制和调度联合设计。考虑基站通过时变信道从多个传感器采集状态信息，以信息年龄最小化为目标，研究满足带宽约束的传输策略。面对多传感器调度问题中的“维度诅咒”，使用带宽松弛和拉格朗日乘子法进行多传感器解耦，设计了松弛约束下的最优策略，并基于此提出了满足带宽约束的调度策略。理论分析证明，所提策略在大规模网络用户数量趋于无穷时有渐近最优性。3. 最小化用户索取状态信息副本的信息年龄的基站端缓存更新。考虑基站端缓存与远端服务器之间的链路带宽受限，针对状态信息流行度时变和下载所需时长时变的特点，通过线性规划和分支定界法设计了基站端缓存更新策略，有效降低了用户从基站端缓存得到的状态信息副本的信息年龄。4. 同步年龄最小化的广播与推送。针对状态信息变化与传输存在双重随机性，以最小化全网用户平均同步年龄为目标，基于怀特指数提出了一种广播推送方案，仿真结果发现所提算法得到的平均同步年龄接近理论下界。

The safety and efficient control of autonomous driving systems relies on status information such as driving speed and highway conditions. These status information are time-sensitive, and are observed by sensors deployed on cars and the surrounding environment. The timeliness of status information requires the communication networks collect status updates from sensors, and send them out to interested users in a timely manner. However, the delivery of fresh status to users is limited by the random transmission delay, the limited communication rate and the time-varying channel conditions due to vehicular mobility. Moreover, data freshness requirement is different from traditional quality of service (QoS) optimization goals such as high throughput and low delay. Thus, in order to guarantee the safety and efficient control of real time applications, it is important to redesign transmission strategies and network scheduling algorithms to obtain a good data freshness performance. This thesis focuses on two scenarios in vehicular networks: the base station collecting status updates from sensors and sending out status updates to interested users. The goal is to propose sampling, transmission and scheduling algorithms that optimize data freshness performance. Depending on the evolution characteristics of status information, two recently proposed metrics, the Age of Information (AoI) and the Age of Synchronization (AoS) are used to measure the freshness of information at the receiver. The main contributions are as follows: 1. Adaptive sampling and transmission strategies in point-to-point communication systems for AoI minimization. Consider a sensor samples status update and transmit them to the destination through a channel with random delay. To overcome the challenge of unknown delay statistics, we propose an online sampling strategy based on Monro-Robbins algorithm. Theoretic analysis proves that the proposed algorithm adaptively learns the optimal sampling policy. The average AoI optimality gap diminishes when the number of samples goes to infinity. 2. AoI optimal sampling, cross-layer transmission and scheduling for multi-sensor networks with time-varying channels. Consider that multiple sensors are connected to the base station via time-varying communication links using a shared bandwidth, the goal is to design scheduling policy that meets the bandwidth constraint of the network and average power constraint of the sensors. To overcome ``the curse of dimension'' in this decision making problem, a sampling and cross-layer scheduling algorithm is proposed through bandwidth relaxation and sensor decoupling. Theoretic analysis shows the asymptotic performance of the algorithm in large scale networks. 3. Caching update strategies for AoI minimization. Consider copies of status information are stored at the base station. When users request for status updates, they obtain the possible outdated copies at the local cache. The goal is to design cache updating strategies for the base station so that the AoI of user requested files can be minimized. AoI minimum cache updating strategies under time-varying files popularity, time-varying update durations are proposed by using the Branch-Reduce-Bound (BRB) and Linear Programming (LP). Simulation results validate the effectiveness of the proposed algorithm in terms of AoI reduction. 4. Data freshness oriented broadcasting in multi-user networks. Consider both status updates generation and transmission encounters randomness. The goal is to minimize the expected average Age of Synchronization (AoS) performance over the entire network by designing efficient broadcasting strategy for the base station. The AoS minimization broadcasting problem is reformulated into a restless multi-arm bandit problem and broadcasting strategy based on the Whittle's index is proposed. Simulation results show the average AoS performance of the proposed algorithm is close to the theoretic lower bound.