In the upcoming sixth-generation (6G) networks, semantic communication has made remarkable strides, where the transceivers utilizing local knowledge bases (KBs) to encode and recover semantic information. In this paper, we propose a universal weighted-KB (UW-KB) endowed with a sample confidence function for an end-to-end (E2E) task-unaware semantic communication system, where both the KB and semantic coding networks at the transceivers are incomplete in the initial stages. This intelligent UW-KB is shaped by receiver feedback during training, autonomously assigning weights to samples to mitigate biases in KB data, which significantly improves the efficiency of semantic coding networks. Simulation results demonstrate the effectiveness of our UW-KB in addressing KB data bias, providing valuable insights to bolster the robustness of task-unaware semantic communication systems.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Universal Weighted-Knowledge Bases for Task-Unaware Semantic Communication Systems


    Contributors:
    Jiang, Shiyao (author) / Jiao, Jian (author) / Zhang, Ke (author) / Wang, Ye (author) / Lu, Rongxing (author) / Zhang, Qinyu (author)


    Publication date :

    2024-06-24


    Size :

    1443132 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English