We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.IT

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Information Theory

Title: Resource Allocation for Uplink Cell-Free Massive MIMO enabled URLLC in a Smart Factory

Abstract: Smart factories need to support the simultaneous communication of multiple industrial Internet-of-Things (IIoT) devices with ultra-reliability and low-latency communication (URLLC). Meanwhile, short packet transmission for IIoT applications incurs performance loss compared to traditional long packet transmission for human-to-human communications. On the other hand, cell-free massive multiple-input and multiple-output (CF mMIMO) technology can provide uniform services for all devices by deploying distributed access points (APs). In this paper, we adopt CF mMIMO to support URLLC in a smart factory. Specifically, we first derive the lower bound (LB) on achievable uplink data rate under the finite blocklength (FBL) with imperfect channel state information (CSI) for both maximum-ratio combining (MRC) and full-pilot zero-forcing (FZF) decoders. \textcolor{black}{The derived LB rates based on the MRC case have the same trends as the ergodic rate, while LB rates using the FZF decoder tightly match the ergodic rates}, which means that resource allocation can be performed based on the LB data rate rather the exact ergodic data rate under FBL. The \textcolor{black}{log-function method} and successive convex approximation (SCA) are then used to approximately transform the non-convex weighted sum rate problem into a series of geometric program (GP) problems, and an iterative algorithm is proposed to jointly optimize the pilot and payload power allocation. Simulation results demonstrate that CF mMIMO significantly improves the average weighted sum rate (AWSR) compared to centralized mMIMO. An interesting observation is that increasing the number of devices improves the AWSR for CF mMIMO whilst the AWSR remains relatively constant for centralized mMIMO.
Comments: Accepted by Transactions on Communications
Subjects: Information Theory (cs.IT); Signal Processing (eess.SP)
Cite as: arXiv:2211.12354 [cs.IT]
  (or arXiv:2211.12354v1 [cs.IT] for this version)

Submission history

From: Qihao Peng [view email]
[v1] Tue, 22 Nov 2022 15:44:18 GMT (5303kb,D)

Link back to: arXiv, form interface, contact.