Distributed AI training often suffers from network faults. Network faults, especially at the last hop between a switch and a host, result in loss of connectivity, resulting in training job stalls and eventual failure. This is typically managed through a fail-stop mechanism, followed by a restart, incurring significant inefficiencies. We present ReCCL, the first network fault-tolerant collective communication library (CCL) that allows training progress to be preserved by seamlessly failing over to alternate paths when a network fault occurs. During failover, ReCCL keeps communication states synchronized while using dynamic channel load balancing and intra-host GPU routing to improve communication performance. Our evaluations demonstrate that ReCCL can perform failover seamlessly with minimal performance losses. Additionally, our simulations also demonstrate that failover can be effectively used to achieve significant savings in GPU hours for large-scale distributed AI training workloads.

Summary Nowadays, wireless radio signals are ubiquitous and are around us; some signals pass through us, and some reflect off us. Substantial advancements in recent years demonstrate that such signals are utilized for diverse emerging applications, including people activity, motion watches, healthcare, and so forth. A few questions would be that may raise severe concerns in future cybersecurity and private domains. For example, what if Wi‐Fi signals are utilized to watch a person doings and actions, which are mostly without the person's authorization and authentication. How far such signal utilization can attack privacy intrusively, silently, more particularly, what/where we do, say, command, see, write, draw, go, perform, everything can be known. In this article, we investigate watching human activities by leveraging Wi‐Fi signals and discuss a few application prototypes. We attempt to learn whether or not attackers have the ability to passively watch our Internet activity as well as physical activities and motions through Wi‐Fi. With all‐new advances, one must be aware that cyberattackers may apply unauthorized use of these advances to their benefit. We discuss some of the countermeasures and approaches to mitigate these risks with Wi‐Fi signal leveraging.

Large-scale Cranes often need to run in a long-term working time, and result in various potential structural health risks. To reduce these risks, we herein design structural health monitoring system of the crane (SHMC for short) with Narrow Band IoT (NB-IoT). Firstly, we design a monitoring terminal for the SHMC system, wherethe single-chip microcomputer controls the electronic device to collect external signals to monitor the crane information in real time. Secondly, we employ NB-IoT based the wireless transmission module for network bandwidth-saving and long-term running. Then, we design a core platform, which parses and calculates the information uploaded by the terminal, and stores it in the database. Our system can monitor the running status of cranes in real time, and timely diagnose the fault information of crane to ensure their safe operation. Also the implementation of our design with apache storm cluster, Kafka server and Alibaba EC2 web server is showed in this study.

We herein build a smart parking system for large-scale underground parking lot in China, and deal with scheduling problems for multiple automatic guided vehicles (AGVs). In previous studies, multiple AGVs usually were scheduled in warehouse system, workshop system, etc. There was little report about AGVs in smart parking. In underground parking lot, designing maximum layout of parking space led to limited and double pathway for AGVs. Therefore, we proposed a dynamic time-window based on scheduling method (DTWS for short). In our DTWS, each AGV can dynamically adjust the path according to a time window, and thus implement the conflict-free pathway of AGVs. Finally, the experimental results showed that the our method solved the scheduling problems such as deadlock of multiple AGVs, and our parking system had an improved performance.