In the sixth generation (6G), ultra-reliable low-latency communications (URLLC) will be further developed to achieve TKμ extreme connectivity. On the premise of ensuring the same rate and reliability, the spatial domain advantage of multiple-input multiple-output (MIMO) has the potential to further shorten the time-domain code length and is expected to be a key enabler for the realization of TKμ. Different coded MIMO schemes exhibit disparities in exploiting the spatial domain characteristics, so we consider two extreme MIMO coding schemes, namely, time-domain coding in which the codewords on multiple spatial channels are independent of each other, and spatiotemporal coding in which multiple spatial channels are jointly coded. By analyzing the statistical characteristics of information density and utilizing the normal approximation, we provide explicit performance bounds for finite blocklength coded MIMO under time-domain coding and spatiotemporal coding. It is found that, different from the phenomenon in time-domain coding where the performance degrades as the blocklengths decrease, spatiotemporal coding can effectively compensate for the performance loss caused by short blocklengths by improving the spatial degrees of freedom (DoF). These results indicate that spatiotemporal coding can fully exploit the spatial dimension advantages of MIMO systems, enabling extremely low error-rate communication under stringent blocklengths constraint.

In order to cope with the demand of explosively increasing service diversity and quality, network slicing has become the key technology of next-generation mobile communication. Mobile edge computing (MEC) can provide multi-dimensional resources and network functions at the edge of the network and reduce the delay of wireless networks. At the same time, non-orthogonal multiple access (NOMA) allows traffic to share resources and improve the spectral efficiency and energy efficiency of wireless networks. In this paper, we propose a hierarchical NOMA-based network slicing architecture in the 6G novel full-spectrum scalable cell-free radio access network with MECs and conduct joint allocation of communication, computing and caching resources at different resource granularity to meet the requirements of latency-critical applications with different latency. In order to realize the hierarchical joint resource allocation to improve system efficiency, we propose to address the optimal computing resource allocation and cache placement problem firstly through conventional optimization methods to reduce the action space and then use the multi-agent deep reinforcement learning algorithm for solving other complex coupling strategies. Simulation results further verify the effectiveness of the proposed intelligent network slicing scheme.

The key performance indicators (KPIs) of the sixth generation (6G) will increase by orders of magnitude compared to the fifth generation (5G), promising extreme connectivity performance with Tbps-scale data rate, Kbps/Hz-scale spectral efficiency (SE) and$\mu \text {s}$-level latency. Cell-free massive MIMO (CF-mMIMO) with rich spatial dimension resources is expected to be a key architecture to realize$\text {TK}\mu $extreme connectivity, but the existing research has not yet given a compact and closed-form approximation to describe the relationship between the spatial dimension and system performance, which makes it difficult to evaluate the KPIs of$\text {TK}\mu $intuitively. This paper derives explicit closed-form expressions for the relationship between system performance and system configuration parameters for finite blocklength CF-mMIMO systems and analyzes the relationship between system performance and spatial dimensions. Based on this, we perform parameter selection and performance evaluation of specific implementations in the three$\text {TK}\mu $KPIs in CF-mMIMO systems. Both theoretical analysis and simulation results show that increasing the spatial degree of freedom (DoF) and deploying antennas more dispersedly can realize latency reduction while guaranteeing the system performance, and the joint collaboration of multi-users and multiple access points (APs) with large DoFs can achieve a continuous increase in SE and data rate.