With growing client diversity, model-heterogeneous personalized federated learning (MHPFL) supports collaboration over structure-heterogeneous client models. However, existing MHPFL methods only achieve client-level personalization but ignore inherent discrepancies within each client's different data samples, leading to limited model performance. To this end, we propose a novel model-heterogeneouspersonalizedFederated learning withMixtureofExperts (pFedMoE) to achieve a fine-grained data-level personalization. As the first work that incorporates MoE in MHPFL, it introduces three innovations: (1) Different clients hold heterogeneous local models, we add a small proxy global homogeneous feature extractor shared by clients for knowledge exchange. (2) To achieve a fine-grained data-level personalization, we construct a personalized local MoE for each client: a local expert (local heterogeneous client model's feature extractor), a global expert (global proxy homogeneous feature extractor), and a local personalized gating network, which dynamically balances the generalization and personalization of the local model at the data sample level. (3) We customize a lightweight linear gating network to capture the generalized and personalized data characteristics of each local data sample. We theoretically prove its$\mathcal {O}(1/T)$convergence rate. Experiments on 3 benchmark image datasets, 1 real-world image dataset and 1 real-world text dataset against 9 baselines demonstrate its state-of-the-art model accuracy with up to 2.79% accuracy improvement while saving up to 43.12% computational overheads and keeping satisfactory communication costs.

Federated learning (FL), an emerging distributed machine learning paradigm, utilizes edge decentralized data from multiple edge nodes (clients) to train a shared model under preserved data privacy. Furthermore, model-heterogeneous personalized federated learning (MHPFL) enables FL clients to train structurally different personalized models on non-independent and identically distributed (non-lID) local data. Existing MHPFL methods focus on data distribution differences among clients, and they propose various client-level personalization approaches to alleviate non-lID issues. However, different data samples in one client may also have different features, which are often ignored, resulting in constrained model performances. To bridge this gap, we propose a novel model-heterogeneous personalized Federated learning approach with Adaptive Feature Mixture (pFedAFM) to achieve data-level personalization while maintaining efficient communication and computation. It consists of three innovative designs: 1) We add a homogeneous small feature extractor alongside each client's local heterogeneous model, and the server aggregates these homogeneous small feature extractors for cross-client knowledge fusion. 2) We design an iterative training strategy to alternately train the global homogeneous small feature extractor and the local heterogeneous client model, for effective bidirectional exchange between global generalized knowledge and local personalized knowledge. 3) During model training, we devise a trainable weight vector to adaptively mix the features (representation) extracted by the global homogeneous and local heterogeneous models for different data samples, i.e., fulfilling data-level personalized feature mixture. Theoretical analysis proves that pFedAFM converges over time. Extensive experiments on 3 computer vision (CV) and 1 natural lan-guage processing (NLP) benchmark datasets demonstrate that pFedAFM significantly outperforms 8 state-of-the-art MHPFL methods, achieving up to 7.93% accuracy improvement while incurring low communication and computation costs.

Learning-based lossless compressors play a crucial role in large-scale genomic database backup, storage, transmission, and management. However, their 1) inadequate compression ratio, 2) low compression & decompression throughput, and 3) poor compression robustness limit their widespread adoption and application in both industry and academia. To solve those challenges, we propose a novel Parallel Multi-Knowledge Learning-based Compressor (PMKLC) with four crucial designs: 1) We propose an automated multi-knowledge learning-based compression framework as compressors' backbone to enhance compression ratio and robustness; 2) we design a GPU-accelerated (s,k)-mer encoder to optimize compression throughput and computing resource usage; 3) we introduce data block partitioning and Step-wise Model Passing (SMP) mechanisms for parallel acceleration; 4) We design two compression modes PMKLC-S and PMKLC-M to meet the complex application scenarios, where the former runs on a resource-constrained single GPU and the latter is multi-GPU accelerated. We benchmark PMKLC-S/M and 14 baselines (7 traditional and 7 leaning-based) on 15 real-world datasets with different species and data sizes. Compared to baselines on the testing datasets, PMKLC-S/M achieve the average compression ratio improvement up to 73.609% and 73.480%, the average throughput improvement up to 3.036X and 10.710X, respectively. Besides, PMKLC-S/M also achieve the best robustness and competitive memory cost, indicating its greater stability against datasets with different probability distribution perturbations, and its strong ability to run on memory-constrained devices. Overall, PMKLC is a balanced compression solution that optimizes compression ratio, throughput, robustness, and resource consumption. PMKLC and linkages of datasets are available at https://github.com/dingyanfeng/PMKLC.

With the rapid development of the Internet, the huge amount of Multi-Source Data (MSD) brings challenges in data sharing and storing. Lossless data compression is the major way to solve those problems. Nowadays, neural-network technologies bring significant advantage in data modeling, making learning-based lossless compressors (LLCs) for multi-source data have emerged continuously. Compared with traditional compressors, the LLCs are more useful to catch complex redundancy patterns in MSD, and thus have great potential in enhancing compression ratio. However, existing LLCs still suffer from unsatisfactory compression ratios and lower throughput. To solve those problems, we propose a novel universal MSD lossless compressor called MSDZip via Stepwise-parallel and learning-based prediction technologies, it introduces two major designs: 1) We propose a Local-Global-Deep Mixing block in the learning-based prediction module to establish dependencies for MSD symbols, where designed Deep Mixing block solves the problem of unstable weights in the perceptual layers caused by cold-start problem to enhance the compression ratio significantly. 2) We design a Stepwise-parallel multi-GPU-accelerated compression strategy to address the compression speed and graphics memory constraints of single GPU in the face of large-scale data. The Stepwise-parallel module passes the source MSD to learning-based prediction model through the data chunking strategy, where the model of the previous chunk is used to guide the compression of the next chunk in parallel. We compare MSDZip with 5 classical learning-based and 6 traditional compressors on 12 well-studied real-world datasets. The experimental results demonstrate that MSDZip optimizes 3.418%-69.874% in terms of compression ratio and 31.171%-495.649% in terms of throughput compared to advanced LLCs. The source code of MSDZip and the linkages of the experimental datasets are available at https://github.com/mhuidong/MSDZip.

Approximate Nearest Neighbor Search (ANNS) has become a fundamental component in many real-world applications. Among various ANNS algorithms, graph-based methods are state-of-the-art. However, ANNS often suffers from a significant drop in accuracy for certain queries, especially in Out-of-Distribution (OOD) scenarios. To address this issue, a recent approach named RoarGraph constructs a bipartite graph between the base data and historical queries to bridge the gap between two different distributions. However, it suffers from some limitations: (1) Building a bipartite graph between two distributions lacks theoretical support, resulting in the query distribution not being effectively utilized by the graph index. (2) Requires a sufficient number of historical queries before graph construction and suffers from high construction times. (3) When the query workload changes, it requires reconstruction to maintain high search accuracy. In this paper, we first propose Escape Hardness, a metric to evaluate the quality of the graph structure around the query. Then we divide the graph search into two stages and dynamically identify and fix defective graph regions in each stage based on Escape Hardness. (1) From the entry point to the vicinity of the query. We propose R eachability Fix ing (RFix), which enhances the navigability of some key nodes. (2) Searching within the vicinity of the query. We propose N eighboring G raph Defects Fix ing (NGFix) to improve graph connectivity in regions where queries are densely distributed. The results of extensive experiments show that our method outperforms other state-of-the-art methods on real-world datasets, achieving up to 2.25× faster search speed for OOD queries at 99% recall compared with RoarGraph and 6.88× faster speed compared with HNSW. It also accelerates index construction by 2.35-9.02× compared to RoarGraph.

The advancement of long reads sequencing technologies has led to a significant increase in biological sequencing big data. Although several reference-free compressors are available for saving long reads data storage space, choosing the suitable one is challenging due to the shortage of thorough and systematic evaluations of their lossless compression effectiveness, both dedicated and general-purpose. In this study, we performed benchmark examinations on 30 compressors, including 11 specialized for long reads and 19 general-purpose ones, using 31 real-world datasets with differing sequencing platforms, species, and lengths. Each lossless compressor was evaluated on 13 performance measures, including compression strength, compression robustness, as well as time and peak memory required for compression and decompression. Additionally, for future long reads data compressors, we outlined investigation directions with consideration for privacy-sensitive sequences data security, hardware parallel acceleration, parameter tuning framework, and system hardware-algorithm integration design. We summarized the results as the Long Reads Compression Benchmark, available at https://github.com/fahaihi/LRCB.

Vector data management systems (VDMSs) have become an indispensable cornerstone in large-scale information retrieval and machine learning systems like large language models. To enhance the efficiency and flexibility of similarity search, VDMS exposes many tunable index parameters and system parameters for users to specify. However, due to the inherent characteristics of VDMS, automatic performance tuning for VDMS faces several critical challenges, which cannot be well addressed by the existing auto-tuning methods. In this paper, we introduce VDTuner, a learning-based automatic performance tuning framework for VDMS, leveraging multi-objective Bayesian optimization. VDTuner overcomes the challenges associated with VDMS by efficiently exploring a complex multi-dimensional parameter space without requiring any prior knowledge. Moreover, it is able to achieve a good balance between search speed and recall rate, delivering an optimal configuration. Extensive evaluations demonstrate that VDTuner can markedly improve VDMS performance (14.12% in search speed and 186.38 % in recall rate) compared with default setting, and is more efficient compared with state-of-the-art baselines (up to 3.57 x faster in terms of tuning time). In addition, VDTuner is scalable to specific user preference and cost-aware optimization objective. VDTuner is available online at https://github.com/tiannuo-yanWVDTuner.

The emerging persistent memory (PM) is increasingly being leveraged to construct high-performance and persistent indexes. By exploiting data distribution, recent learned indexes open up a new index design paradigm. Some prior studies try to refit the learned index according to the features of PM. However, they neglect to analyze the performance of existing learned index schemes on PM. In this paper, we provide a comprehensive analysis of learned indexes on PM and propose two optimization methods to improve the performance. In particular, we evaluate ALEX, PGM-index, and XIndex after converting them to persistent indexes. With appropriate modifications, some design choices of volatile learned index still show favorable performance on PM under workloads with simple data distribution. But they perform poorly when the data distribution becomes complex. According to the experiment results, we summarize some instructive insights and optimize persistent learned indexes for complex data distributions with two methods: 1) a cost-based insertion pattern selection to minimize PM writes and 2) recoverable internal nodes selective persistence to decrease the overhead of internal lookups. Our evaluations demonstrate the performance of optimized ALEX is 2.09x/1.53x of the original ALEX in insert/search. Meanwhile, it also outperforms the specific-designed persistent learned index.

Time series databases play a critical role in large-scale performance monitoring systems. Metrics are required to be observable immediately after being generated to support real-time analysis. However, the commonly used Log-Structured Merge-Tree structure suffers from periodically visible delay spikes when a new segment is created due to the instantaneous index construction pressure.

In-memory Key-Value (KV) stores are widely deployed in modern data centers. These systems generally use DRAM as their storage medium, causing huge hardware costs. The emerging persistent memory (PMEM) is a potential substitute for DRAM, which has a lower price and larger capacity, but lower access speed and bandwidth. Many prior studies strive to build hybrid memory systems to retain both the advantages of DRAM and PMEM. However, they are either application agnostic or simply take DRAM as a cache, which are both not efficient for in-memory KV stores. In this paper, we propose SwapKV, a well-designed in-memory KV store for hybrid DRAM-PMEM system. SwapKV has several promising properties. First, SwapKV combines DRAM and PMEM to a uniform memory pool and only stores one copy of data, which maximizes capacity utilization. Second, SwapKV maps all writing operations to DRAM and migrates data to PMEM with large blocks asynchronously, which mitigates the intrinsic inefficiency of PMEM for writing operations. Third, SwapKV maintains the hot data in DRAM through an efficient hotness filtering and data swapping mechanism, which ensures high system throughput and responsiveness. We implement SwapKV and evaluate it under various workload patterns. The results demonstrate that SwapKV improves the throughput by 11\%$\sim$41\% compared to the state-of-the-art alternatives.

Commercial search engines generally maintain hundreds of thousands of machines equipped with large sized DRAM which incurs high hardware cost since DRAM is expensive. Recently, NVM Optane SSD has been considered as a promising underlying storage device due to its price advantage and speed advantage. However, to achieve a comparable efficiency performance with in-memory index, applying NVM to both latency and I/O bandwidth critical applications still face non-trivial challenges, because NVM has much lower I/O speed and bandwidth compared to DRAM. In this paper, we propose an NVM SSD-optimized query processing framework, aiming to address both the latency and bandwidth issues of using NVM in search engines. First, we propose a pipelined query processing methodology which significantly reduces the I/O waiting time. Second, we propose a cache-aware query reordering algorithm which enables queries sharing more data to be processed adjacently. Third, we propose a data prefetching mechanism which reduces the extra thread waiting time and improves bandwidth utilization. Moreover, we propose intra-query parallel mechanisms for long-tail queries, including query subtask scheduling, heap concurrent access strategy, query parallelism prediction and adaptive pipelining. Extensive experimental studies show that our framework significantly outperforms the state-of-the-art baselines, which obtains comparable processing latency and throughput with DRAM in both inter-query and intra-query parallel scenarios.

General anomaly detection techniques have always received a lot of attention. Current detection methods usually focus solely on representation learning or anomaly judgment. This paper proposes a Multi-Subspace Deviation Network (MSDN) framework to build a model combining feature learning with anomaly score learning under the condition that a small number of labeled anomalies can be observed. Concretely, our framework combines a feature learner with two specific projectors: a self-supervised projector and an anomaly score learner. We utilize random affine transformations to map the raw data to multiple subspaces and train a classifier to predict the transformation label in the self-supervised module. Anomaly scores are then obtained directly from a deviation network, where the contrastive loss is used to amplify the gap in the anomaly scores between normal objects and anomalies. Extensive experiments on eight datasets show that our proposed method achieves higher detection accuracy than previous schemes with fewer observed anomalies.

Existing online knowledge distillation (KD) methods solve the dependency problem of the high-capacity teacher model via mutual learning and ensemble learning. But they focus on the utilization of logits information in the last few layers and fail to construct a strong teacher model to better supervise student networks, leading to the inefficiency of KD. In this work, we propose a simple but effective online knowledge distillation algorithm, called Attentive Feature Interaction Distillation (AFID). It applies interactive teaching in which the teacher and the student can send, receive, and give feedback on an equal footing, ultimately promoting the generality of both. Specifically, we set up a Feature Interaction Module for two sub-networks to conduct low-level and mid-level feature learning. They can alternately transfer attentive features maps to exchange interesting regions and fuse the other party’s map with the features of self-extraction for information enhancement. Besides, we assign a Feature Fusion Module, in which a Peer Fused Teacher is formed to fuse the output features of two sub-networks to guide sub-networks and a Peer Ensemble Teacher is established to accomplish mutual learning between the two teachers. Integrating Feature Interaction Module and Feature Fusion Module into a unified framework takes full advantage of the interactive teaching mechanism and makes the two sub-networks capture and transfer more fine-grained features to each other. Experimental results on CIFAR-100 and ImageNet ILSVRC 2012 real datasets show that AFID achieves significant performance improvements compared with existing online KD and classical teacher-guide methods.

The performance of query processing has always been a concern in the field of information retrieval. Dynamic pruning algorithms have been proposed to improve query processing performance in terms of efficiency and effectiveness. However, a single pruning algorithm generally does not have both advantages. In this work, we investigate the performance of the main dynamic pruning algorithms in terms of average and tail latency as well as the accuracy of query results, and find that they are complementary. Inspired by these findings, we propose two types of hybrid dynamic pruning algorithms that choose different combinations of strategies according to the characteristics of each query. Experimental results demonstrate that our proposed methods yield a good balance between both efficiency and effectiveness.

Commercial search engines generally maintain hundreds of thousands of machines equipped with large sized DRAM in order to process huge volume of user queries with fast responsiveness, which incurs high hardware cost since DRAM is very expensive. Recently, NVM Optane SSD has been considered as a promising underlying storage device due to its price advantage over DRAM and speed advantage over traditional slow block devices. However, to achieve a comparable efficiency performance with in-memory index, applying NVM to both latency and I/O bandwidth critical applications such as search engine still faces non-trivial challenges, because NVM has much lower I/O speed and bandwidth compared to DRAM.

In query processing, incorporating proximity between query terms is beneficial for effective retrieval. However, it brings inevitable storage and computing costs by using positional data in inverted indexes. In this paper, we propose a lossy method for compressing term position data in the case of utilizing term proximity. Our method exploits clustering property of term occurrences, adaptively clusters the nearby occurrences, and replaces the clustered positions with a centralized value. Experimental results show that our adaptive method is competitive with respect to index size, ranking efficiency and effectiveness.

Search engines encounter a time vs. space trade-off: search responsiveness (i.e., a short query response time) comes at the cost of increased index storage. We propose a hybrid method which uses both (a) the recently published mapping-matrix-style index BitFunnel (BF) for search efficiency, and (b) the state-of-the-art Partitioned Elias-Fano (PEF) inverted-index compression method. We use this proposed hybrid method to minimize time while satisfying a fixed space constraint, and to minimize space while satisfying a fixed time constraint. Each document is stored using either BF or PEF, and we use a local search strategy to find an approximately optimal BF-PEF partition. Since performing full experiments on each candidate BF-PEF partition is impractically slow, we use a regression model to predict the time and space costs resulting from candidate partitions (space accuracy 97.6%; time accuracy 95.2%). Compared with a hybrid mathematical index (Ottaviano et al., 2015), the time cost is reduced by up to 47% without significantly exceeding its size. Compared with three mathematical encoding methods, the hybrid BF-PEF index allows performing list intersection between around 16% to 76% faster (without significantly increasing the index size). Compared with BF, the index size is reduced by 45% while maintaining an intersection time comparable to that of BF.

Large-scale search engines utilize inverted indexes which store ordered lists of document identifies (docIDs) relevant to query terms, which can be queried thousands of times per second. In order to reduce storage requirements, we propose a dictionary-based compression approach for the recently proposed bitwise data-structure BitFunnel, which makes use of a Bloom filter. Compression is achieved through storing frequently occurring blocks in a dictionary. Infrequently occurring blocks (those which are not represented in the dictionary) are instead referenced using similar blocks that are in the dictionary, introducing additional false positive errors. We further introduce a docID reordering strategy to improve compression. Experimental results indicate an improvement in compression by 27% to 30%, at the expense of increasing the query processing time by 16% to 48% and increasing the false positive rate by around 7.6 to 10.7 percentage points.

Large-scale search engines need to answer thousands of queries per second over billions of documents, which is typically done by querying a large inverted index. Many highly optimized integer encoding techniques are applied to compress the inverted index and reduce the query processing time. In this paper, we propose a new grammar-based inverted index compression scheme, which can improve the performance of both index compression and query processing.

In response to a user query, search engines return the top-k relevant results, each of which contains a small piece of text, called a snippet, extracted from the corresponding document. Obtaining a snippet is time consuming as it requires both document retrieval (disk access) and string matching (CPU computation), so caching of snippets is used to reduce latency. With the trend of using flash-based solid state drives (SSDs) instead of hard disk drives for search engine storage, the bottleneck of snippet generation shifts from I/O to computation. We propose a simple, but effective method for exploiting this trend, which we call fragment caching: instead of caching the whole snippet, we only cache snippet metadata which describe how to retrieve the snippet from the document. While this approach increases I/O time, the cost is insignificant on SSDs. The major benefit of fragment caching is the ability to cache the same snippets (without loss of quality) while only using a fraction of the memory the traditional method requires. In our experiments, we find around 10 times less memory is required to achieve comparable snippet generation times for dynamic memory, and we consistently achieve a vastly greater hit ratio for static caching.

Representing words into vectors in continuous space can form up a potentially powerful basis to generate high-quality textual features for many text mining and natural language processing tasks. Some recent efforts, such as the skip-gram model, have attempted to learn word representations that can capture both syntactic and semantic information among text corpus. However, they still lack the capability of encoding the properties of words and the complex relationships among words very well, since text itself often contains incomplete and ambiguous information. Fortunately, knowledge graphs provide a golden mine for enhancing the quality of learned word representations. In particular, a knowledge graph, usually composed by entities (words, phrases, etc.), relations between entities, and some corresponding meta information, can supply invaluable relational knowledge that encodes the relationship between entities as well as categorical knowledge that encodes the attributes or properties of entities. Hence, in this paper, we introduce a novel framework called RC-NET to leverage both the relational and categorical knowledge to produce word representations of higher quality. Specifically, we build the relational knowledge and the categorical knowledge into two separate regularization functions, and combine both of them with the original objective function of the skip-gram model. By solving this combined optimization problem using back propagation neural networks, we can obtain word representations enhanced by the knowledge graph. Experiments on popular text mining and natural language processing tasks, including analogical reasoning, word similarity, and topic prediction, have all demonstrated that our model can significantly improve the quality of word representations.

Caching is an important optimization in search engine architectures. Existing caching techniques for search engine optimization are mostly biased towards the reduction of random accesses to disks, because random accesses are known to be much more expensive than sequential accesses in traditional magnetic hard disk drive (HDD). Recently, solid-state drive (SSD) has emerged as a new kind of secondary storage medium, and some search engines like Baidu have already used SSD to completely replace HDD in their infrastructure. One notable property of SSD is that its random access latency is comparable to its sequential access latency. Therefore, the use of SSDs to replace HDDs in a search engine infrastructure may void the cache management of existing search engines. In this article, we carry out a series of empirical experiments to study the impact of SSD on search engine cache management. Based on the results, we give insights to practitioners and researchers on how to adapt the infrastructure and caching policies for SSD-based search engines.

Caching is a widely used technique to boost the performance of search engines. Based on the observation that the speed gap between the random access of flash-based solid state drive and its sequential access is much inapparent than that of magnetic hard disk drive, we introduce a new static list caching algorithm which takes the block-level access latency into consideration. The experimental results show that the proposed policy can reduce the average disk access latency per query by up to 14\% over the state-of-the-art algorithms in the SSD-based infrastructure. Besides, the results also reveal that our new strategy outperforms other existing algorithms even on HDD-based architecture.

Caching technologies have been widely employed to boost the performance of Web search engines. Motivated by the correlation between terms in query logs from a commercial search engine, we explore the idea of a caching scheme based on pairs of terms, rather than individual terms (which is the typical approach used by search engines today). We propose an inverted list caching policy, based on the Least Recently Used method, in which the co-occurring correlation between terms in the query stream is accounted for when deciding on which terms to keep in the cache. We consider not only the term co-occurrence within the same query but also the co-occurrence between separate queries. Experimental results show that the proposed approach can improve not only the cache hit ratio but also the overall throughput of the system when compared to existing list caching algorithms.

Caching is an important optimization in search engine architectures. Existing caching techniques for search engine optimization are mostly biased towards the reduction of random accesses to disks, because random accesses are known to be much more expensive than sequential accesses in traditional magnetic hard disk drive (HDD). Recently, solid state drive (SSD) has emerged as a new kind of secondary storage medium, and some search engines like Baidu have already used SSD to completely replace HDD in their infrastructure. One notable property of SSD is that its random access latency is comparable to its sequential access latency. Therefore, the use of SSDs to replace HDDs in a search engine infrastructure may void the cache management of existing search engines. In this paper, we carry out a series of empirical experiments to study the impact of SSD on search engine cache management. The results give insights to practitioners and researchers on how to adapt the infrastructure and how to redesign the caching policies for SSD-based search engines.

Major web search engines answer thousands of queries per second requesting information about billions of web pages. The data sizes and query loads are growing at an exponential rate. To manage the heavy workload, we consider techniques for utilizing a Graphics Processing Unit (GPU). We investigate new approaches to improve two important operations of search engines -- lists intersection and index compression. For lists intersection, we develop techniques for efficient implementation of the binary search algorithm for parallel computation. We inspect some representative real-world datasets and find that a sufficiently long inverted list has an overall linear rate of increase. Based on this observation, we propose Linear Regression and Hash Segmentation techniques for contracting the search range. For index compression, the traditional d-gap based compression schemata are not well-suited for parallel computation, so we propose a Linear Regression Compression schema which has an inherent parallel structure. We further discuss how to efficiently intersect the compressed lists on a GPU. Our experimental results show significant improvements in the query processing throughput on several datasets.

This paper analyzes churn impact on replicated data duration with different node lifetime distributions. In structured overlay networks, churn includes node-join churn and node-failure churn, caused by the arrival and departure of nodes separately. The paper introduces a duration model of replicated data under node-failure churn for node failure directly leads to data loss. Furthermore, it investigates the impact of node-join churn on the duration of replicated data for different node-lifetime distributions. The paper presents that node-churn will negatively impact on replicated data duration for heavy-tailed distribution and Weibull distribution except exponential distribution. Then we evaluate the impact on replicated data duration with two real-world trace datasets. The experimental results show the negative impact of node-join churn for different node-join churn degrees. Finally, the paper discusses an enhancement by setting a trial period for every fresh node. By experiment, it is an effective way to reduce the negative impact of node-join churn due to the memory property of node lifetime distributions.