In recent years, federated learning (FL) has gained increasing attention in industrial Internet-of-Things (IIoT) domains due to its privacy-preserving advantages. However, prior works commonly adopt a one-size-fits-all strategy for FL computation resource management and reward allocation, disregarding the time-varying participant states across different FL training rounds. Consequently, these methods fail to ensure the sustainability and active participation of IIoT devices in realistic FL deployments. To bridge this gap, we propose a personalized FL methodology for green IIoT systems powered by renewable energy sources. We first establish an incentive model along with its preference parameter-solving scheme to accurately characterize the incentive preferences of individual FL participants. Subsequently, a personalized participant scheduling approach is developed to accommodate dynamic resource usage patterns and diverse incentive preferences among FL participants. Our technique integrates empirical insights into conventional proximal policy optimization methods to accelerate policy learning within reinforcement learning frameworks. Experimental results on an FL prototype system show that our methodology improves the FL model accuracy by 25.92% compared with representative baseline algorithms.

Mobile devices have become ubiquitous in daily life. In contrast to traditional servers, mobile devices suffer from limited memory resources, leading to a significant degradation in the user experience. This paper demonstrates that the primary cause of memory consumption lies in anonymous pages associated with application heaps. Existing schemes are ineffective in deduplicating these pages due to the limited occurrence of the same anonymous pages. This paper presents Sparrow, a similar-page aware deduplication solution for mobile systems. Sparrow shows that memory pages still have the potential to deduplicate, even though the same pages are rare. An interesting observation inspires this, that is, a high number of pages having the partially-same contents. We have implemented Sparrow on real-life smartphones. Experimental results indicate that 30.45% more space can be saved with Sparrow.

Racetrack memory (RTM), which consists of multiple domain block clusters (DBC) and access ports, is a novel non-volatile memory and has potential as scratchpad memory (SPM) in embedded devices due to its high density and low access latency. However, too many shift operations decrease the performance of RTM and cause unpredictable performance. In this paper, we propose three schemes to optimize the performance of RTM from different aspects, including intra-DBC, inter-DBC, and hybrid SPM with SRAM and RTM. Firstly, a balanced group-based data placement method for the data layout inside one DBC is proposed to reduce shifts. Second, a grouping method for the data allocation among DBCs is proposed. It helps with the shift reduction while using fewer DBCs by using one DBC as multiple DBCs. Finally, we use SRAM to further help the cost reduction, and a cost evaluation metric is proposed to assist the shrinking method which determines the data allocation for hybrid SPM with SRAM and RTM. Experiments show that the proposed schemes can significantly improve the performance of pure RTM and hybrid SPM while using fewer DBCs.

Nonvolatile memory (NVM) has the potential as the medium for scratchpad memory (SPM) in embedded devices. Racetrack memory (RM), in particular, is a developing memory technology that possesses high density and read latency comparable to SRAM. The RM’s access operations, however, are based on shift operations. Multiple shift operations will lead to long access latency and high energy. In this article, SRAM is borrowed to help the shifts reduction. Thus, a novel hybrid SRAM+RM SPM is presented to make use of SRAM’s random access and RM’s high density. But, there are some challenges to the proposed architecture: 1) the large capacity of SRAM is not available due to its low density and 2) due to the drawbacks of RM mentioned above, data that are randomly accessed are not expected to be stored on RM. Therefore, a data placement scheme and an instruction scheduling strategy are presented for the proposed architecture. First, an access instruction scheduling strategy is introduced to obtain a relatively sequential access sequence to help with the shifts and SRAM size reduction; second, to help with data placement, a metric for representing the data access cost is proposed; third, a data placement strategy based on the metric is proposed; and finally, a solution for decreasing SRAM size is suggested to maximize the capacity of SPM (or minimize the size of SPM). Experiments show that the suggested scheme can significantly improve the performance of the hybrid SPM while also reducing the shifts on RM with minimal SRAM.

Along with the progress of AI democratization, machine learning (ML) has been successfully applied to edge applications, such as smart phones and automated driving. Nowadays, more applications require ML on tiny devices with extremely limited resources, like implantable cardioverter de-fibrillator (ICD), which is known as TinyML. Unlike ML on the edge, TinyML with a limited energy supply has higher demands on low-power execution. Stochastic computing (SC) using bitstreams for data representation is promising for TinyML since it can perform the fundamental ML operations using simple logical gates, instead of the complicated binary adder and multiplier. However, SC commonly suffers from low accuracy for ML tasks due to low data precision and inaccuracy of arithmetic units. Increasing the length of the bitstream in the existing works can mitigate the precision issue but incur higher latency. In this work, we propose a novel SC architecture, namely Block-based Stochastic Computing (BSC). BSC divides inputs into blocks, such that the latency can be reduced by exploiting high data parallelism. Moreover, optimized arithmetic units and output revision (OUR) scheme are proposed to improve accuracy. On top of it, a global optimization approach is devised to determine the number of blocks, which can make a better latency-power trade-off. Experimental results show that BSC can outperform the existing designs in achieving over 10% higher accuracy on ML tasks and over$6\times$power reduction.

Non-volatile memory (NVM) is expected to be the second level memory (named remote memory) in two-level memory hierarchy in the future. However, NVM has the limited write endurance, thus it is vital to reduce the number of write operations on NVM. Meanwhile, in two-level memory hierarchy, prefetch is widely used for fetching certain data before it is actually required, to hide the remote memory access latency. In general, large-scale nested loop is the performance bottleneck in one program due to the write operations on NVM caused by the first level memory (named local memory) miss and data reuse. Loop tiling is the key technique for grouping iterations so as to reduce the communication with remote memory used in compiler. In this paper, we propose a new loop tiling approach for minimizing the write operations on NVMs and completely hiding the NVM access latency. Specifically, we introduce a series of theorems to help loop tiling. Then, a legal tile shape and an optimal tile size selection strategy is proposed according to data dependency and local memory capacity. Furthermore, we propose a pipeline scheduling policy to completely hide the remote memory latency. Extensive experiments show that the proposed techniques can reduce write operations on NVMs by 95.1% on average, and NVM latency can be completely hidden.

Racetrack memory (RM) has high access performance comparable to SRAM. It is a kind of non-volatile memory (NVM), which consists of data block clusters (DBCs) and access ports. However, data accessing on RM is based on shift operations, which will decrease the performance of RM. This paper proposes techniques by using SRAM to reduce the shifts and improve the accessing performance of RM. The key idea is to place randomly accessed data on SRAM ahead of time to relax the data placement on RM. First, a greedy scheduling strategy is proposed to reduce the requirement of SRAM. Second, to further reduce shifts, data with similar association degree are grouped and allocated to each DBC. Experimental results show that the proposed techniques reduce the shifts by 72.3% with only 256-byte SRAM compared to pure RM.

Efficiently accessing remote file data remains a challenging problem for data processing systems. Development of technologies in non-volatile dual in-line memory modules (NVDIMMs), in-memory file systems, and RDMA networks provide new opportunities towards solving the problem of remote data access. A general understanding about NVDIMMs, such as Intel Optane DC Persistent Memory (DCPM), is that they expand main memory capacity with a cost of multiple times lower performance than DRAM. With an in-depth exploration presented in this paper, however, we show an interesting finding that the potential of NVDIMMs for high-performance, remote in-memory accesses can be revealed through careful design. We explore multiple architectural structures for accessing remote NVDIMMs in a real system using Optane DCPM, and compare the performance of various structures. Experiments are conducted to show significant performance gaps among different ways of using NVDIMMs as memory address space accessible through RDMA interface. Furthermore, we design and implement a prototype of user-level, in-memory file system, RIMFS, in the device DAX mode on Optane DCPM. By comparing against the DAX-supported Linux file system, Ext4-DAX, we show that the performance of remote reads on RIMFS over RDMA is 11.44 higher than that on a remote Ext4-DAX on average. The experimental results also show that the performance of remote accesses on RIMFS is maintained on a heavily loaded data server with CPU utilization as high as 90%, while the performance of remote reads on Ext4-DAX is significantly reduced by 49.3%, and the performance of local reads on Ext4-DAX is even more significantly reduced by 90.1%. The performance comparisons of writes exhibit the same trends.

In this paper, a novel hybrid scratchpad memory (SPM) with SRAM and racetrack memory (RM) is proposed. The basic idea is to smartly place data on SPM by taking the advantages of these two memories. First, a metric is proposed to represent the access cost of data; Second, a data placement scheme is proposed based on the metric; Finally, to maximize the size of SPM, a scheme is further proposed to minimize the size of SRAM. Experimental results show that the proposed scheme reduces the shift operations of RM by 80.12% and reduces the cost of SPM by 80.72% with only 17.63% SRAM compared with a baseline SPM with pure RM.