@phdthesis{
  author = {G. Xu},
  year = {2020},
  month = {Mar},
  title = {In-Network Processing Algorithms for Cooperative Networks},
  publisher = {Shaker Verlag},
  ISBN = {978-3-8440-7270-9},
  address={Bremen, Germany},
  abstract={The dissertation is about distributed algorithms for cooperative processing among collaborated nodes within a network (In-Network Processing (INP)). The main content is mainly focused on the novel INP algorithms development of distributed signal processing that are applied to cooperative networks like the mobile communications networks or wireless sensor networks. For the distributed applications, the INP can generally provide solutions to distributed inference problems based on measurements of physical, medical, biological chemical sensors in wireless sensor networks, in addition, the INP can also facilitate distributed signal processing for distributed radio access networks in 5G/6G communications system.  In this dissertation, a variety of distributed algorithms under the framework of INP have been provided. These INP algorithms are designed for application scenarios where joint processing tasks are executed by cooperated entities in wireless communications networks through a distributed and parallel fashion. In particular, two main areas the distributed consensus-based joint estimation for signal detection and distributed joint precoding for signal transmission in communications networks are highly focused on and investigated in this dissertation.   The INP algorithms for distributed consensus-based joint estimation are applied to the scenario where a group of nodes receive signals from common users, and those receive nodes aim to collaborate to estimate the received signals in a distributed way, such that all receive nodes can eventually achieve same estimates on the user signals. To develop the INP algorithms for distributed consensus-based joint estimation, several numerical approaches like Lagrangian dual method and alternating direction method of multipliers have been investigated in order to solve for corresponding constrained optimization problems in a distributed way. In addition, some novel approaches like the virtual clustering as well as consensus-achieving filtering approaches have also be invented to improve the performance with low communication overhead and low computation complexity during the cooperative processing. Moreover, the INP algorithms for the distributed joint precoding are applied to another scenario where a group of nodes cooperate to perform joint transmission to target users  in order to avoid interference at the receivers. Correspondingly, by exploiting advanced numerical approaches for distributed implementation of joint signal processing like two-step Jacobi and pre-conditioned Richardson methods, several types of INP algorithms for distributed joint precoding with low communication effort and low computation complexity have been developed. Meanwhile, different transmit power constraints are also taken into account for the development of different distributed precoding algorithms to cope with different practical system requirements.  Overall, this work not only presents a variety of concrete INP algorithms for distributed processing in communication networks, it also provides novel ideas and deep analysis of the INP approaches and techniques that can be applied to a wide range of applications for distributed joint processing in cooperative networks.}
}