近年来,随着物联网(Internet of things,IoT)设备的大规模部署,针对物联网设备的恶意代码也不断出现,物联网安全面临来自恶意代码的巨大威胁,亟需对物联网恶意代码检测技术进行综合研究.随着人工智能(artificial intelligence,AI)在计算机视觉和自然语言处理等领域取得了举世瞩目的成就,物联网安全领域也出现了许多基于人工智能的恶意代码检测工作.通过跟进相关研究成果,从物联网环境和设备的特性出发,提出了基于该领域研究主要动机的分类方法,从面向物联网设备限制缓解的恶意代码检测和面向性能提升的物联网恶意代码检测2方面分析该领域的研究发展现状.该分类方法涵盖了物联网恶意代码检测的相关研究,充分体现了物联网设备独有的特性以及当前该领域研究存在的不足.最后通过总结现有研究,深入讨论了 目前基于人工智能的恶意代码检测研究中存在的问题,为该领域未来的研究提出了结合大模型实现物联网恶意代码检测,提高检测模型安全性以及结合零信任架构3个可能的发展方向.

As one of the most important wetland systems, coastal wetlands play an important role in conserving water, regulating the climate and protecting biodiversity. However, due to large-scale and long-term tidal flat reclamations, the landscape structure and function of the coastal wetlands have been greatly affected. Therefore, it is necessary to understand the spatio-temporal characteristics of the impact of tidal flat reclamation on regional ecology and to quantitatively assess the relationships between them. In this study based on long-term, multiperiod remote sensing data, the main spatio-temporal variation characteristics of stability, and the relationship between stability and tidal flat reclamation were analyzed with regard to the influence scope of tidal flat reclamation. The results showed that a substantial decrease in natural wetlands in 1980, mainly caused by tidal flat reclamation, was discovered in the Jiangsu coastal area, and the influence scope of tidal flat reclamation on regional landscape ecology was roughly 30 km. In the affected area, the overall stability had a tendency to improve, but the stability change characteristics between reclamation area and non-reclamation area varied greatly. Especially in the reclamation area, the stability of construction wetlands and non-wetlands deteriorated. Spatially, the stability outside the reclamation area had the characteristics of first deteriorating and then improving as the distance from the reclamation area increased. Under the influence of tidal flat reclamation, the influence of different use types of TFR on stability was not completely consistent, and the influence of the same uses type of tidal flat reclamation on different landscapes was also different.

过去几年中,机器学习算法在计算机视觉、自然语言处理等领域取得了巨大成功.近年来,工业互联网安全领域也涌现出许多基于机器学习技术的入侵检测工作.从工业互联网的自身特性出发,对目前该领域的相关工作进行了深入分析,总结了工业互联网入侵检测技术研究的独特性,并基于该领域中存在的3个主要研究问题提出了新的分类方法,将目前基于机器学习的互联网入侵检测技术分为面向算法设计的研究工作、面向应用限制和挑战的研究工作,以及面向不同ICS攻击场景的研究工作.该分类方法充分展现了不同研究工作的意义以及该领域目前研究工作中存在的问题,为未来的研究工作提供了很好的方向和借鉴.最后基于目前机器学习领域的最新进展,为该领域未来的发展提出了 2个研究方向.

Tidal flats have great potential to expand human living space and provide abundant resources. However, large-scale reclamation and development activities have had profound impacts on regional ecosystems, and it is important to grasp the comprehensive characteristics of the effects from a macro perspective. We selected the coastal area Jiangsu, China, as a typical area, and investigated the impact of reclamation on the coastal wetland system from the perspective of ecosystem services stability. The optimal equilibrium state was used as a benchmark and a stability deviation index (SDI) of ecosystem services was constructed by clarifying the connotation of stability. The stable state and evolution characteristics of ecosystems in different areas (coastal administrative areas, tidal flat areas, and Yancheng National Nature Reserve (YNNR)) were screened and evaluated. Based on the SDI, an overall ecological impact assessment system was developed, and the impact of tidal flat reclamation on ecological stability and its cumulative effects were quantified. The results showed that in the past 40 years, the overall ecological stability of Jiangsu coastal area was “Basically Stable (II)”, while that of the YNNR was “Weakly Stable (III)”. The impact of reclamation on the ecosystem stability of the coastal tidal flat area was significant, with a contribution rate of 73.5 %, and the threshold of stability transition in this area was reached when the reclamation amount was 1251.33 km2. The extent to which reclamation explained the stability variation in the core zone, buffer zone, and experimental zone of YNNR gradually decreased, indicating that the impact of reclamation was greater when strong protection measures were in place. The southern region will reach the maximum threshold of the cumulative effect of instability in 2072. This study provides ideas for the analysis of the comprehensive impact of reclamation on coastal wetland ecosystems, and can provide guidance for the efficient utilization and protection of tidal flats.

在西藏阿里地区狮泉河上游的大、小昂龙冰川开展了连续2年(2014—2016年)的冰川变化地面观测,主要包括冰川表面物质平衡与差分GPS高程变化同步观测,以及冰川表面流速观测,冰川末端观测和冰川雷达测厚.观测结果表明:大、小昂龙冰川表面物质平衡与同期差分GPS观测结果之间存在差异.冰川表面物质平衡结果显示,2014—2016年间,大、小昂龙冰川分别以每年72 mm w.e.和219 mm w.e.的速率减薄.差分GPS观测结果显示,同期大、小昂龙冰川分别以每年(442±90)mm w.e.和(265±90)mm w.e.的速率减薄;在2015/2016年,大、小昂龙冰川表面平均流速分别为4.4 m·a-1和2.3 m·a-1,其中大昂龙冰川表面平均流速较上一物质平衡年增加了10.5％;2014—2016年间,小昂龙冰川先是前进了11 m,之后又退缩了34 m,两年内平均每年退缩11.5 m;大昂龙冰川平均冰厚为67.9 m,实测最大厚度为216 m,根据雷达测厚数据插值计算的冰川储量为0.452 km3;小昂龙冰川实测最大厚度为190 m.

In recent years, machine learning (ML) has become the mainstream network intrusion detection system(NIDS).However, the inherent vulnerabilities of machine learning make it difficult to resist adversarial attacks, which can mislead the models by adding subtle perturbations to the input sample.Adversarial machine learning (AML) has been extensively studied in image recognition.In the field of intrusion detection, which is inherently highly antagonistic, it may directly make ML-based detectors unavailable and cause significant property damage.To deal with such threats, the latest work of applying AML technology was systematically investigated in NIDS from two perspectives: attack and defense.First, the unique constraints and challenges were revealed when applying AML technology in the NIDS field; secondly, a multi-dimensional taxonomy was proposed according to the adversarial attack stage, and current work was compared and summarized on this basis; finally, the future research directions was discussed.

沿海滩涂湿地是脆弱的生态敏感区,滩涂围垦是沿海国家和地区拓展土地资源的重要方式之一.近年,沿海地区快速城市化下的滩涂围垦导致了沿海湿地生态系统退化等一系列环境问题.基于环境库兹涅茨曲线(Environmental Kuznets Curve,EKC)模型,以江苏北部沿海湿地为研究区,结合遥感影像和社会经济数据,刻画滩涂围垦过程、强度及社会经济发展过程,揭示近40年江苏沿海滩涂围垦主要类型、阶段性变化特征,阐述滩涂围垦与社会经济发展的耦合关系.研究结果显示,养殖水体扩张是近40年江苏滩涂围垦利用的主要方式;相比养殖水体聚集发展的盐城市,以启东为典型代表的南通市经历了相对较完整的养殖水体的滩涂围垦过程,其养殖水体面积与区域渔业产值存在指数关系,即当人均地区产值达到约9.55万元时,区域渔业产值的增长不再以通过围垦增加养殖水体面积为主要途径;从江苏沿海滩涂围垦来看,滩涂围垦与社会经济发展之间存在典型倒U型曲线关系.苏北滩涂围垦可分为3个阶段,初期阶段(1980-2005年),滩涂围垦随着经济发展而急剧增加;中期(2005-2014年),当人均GDP约为3.64× 104元时滩涂围垦速度达最大;后期(2014-2018年),滩涂围垦随着经济发展呈现下降趋势.3个阶段中,单位滩涂围垦面积与人均GDP增长关系由4.11元/km2,17.08元/km2增加至60.77元/km2,呈显著增加的趋势.经济发展,农业和农村经济改革以及国家和地方各级政策均对滩涂围垦具有重要影响.严格的国家环境保护政策将是有效控制滩涂围垦、保护生态环境的重要手段.

Competition is an important factor in determining the evolution of ecosystem population (or community) structure and the change in system functions, which are usually manifested directly at the system structure level. From the perspective of ecosystem services, based on the Pareto optimisation theory, we selected the coastal area of Jiangsu, China as the study area and carried out a study of equivalent competition among ecosystem populations (or communities). Firstly, we divided the study area into the region with nature reserve (NRR) and the region without nature reserve (Non-NRR). The study was performed through clarifying the 'structure–process–function–service' cascade relationship of the ecosystem and quantifying the value of major ecosystem services. Then, based on the theory of competition/coexistence, considered the population (or community) in the regional complex ecosystem as the object, we used the reference point-based non-dominated sorting (NSGA-III) algorithm to simultaneously optimise each ecosystem service to obtain a compromised Pareto optimal solution set. Finally, the technique for order preference by similarity to an ideal solution (TOPSIS) was used as a comprehensive evaluation method based on the criteria importance through inter-criteria correlation (CRITIC) weighting method for multi-objective decision-making to obtain the optimal balance of regional ecosystem services. The optimal balance implies that various groups can not only ensure that their own service values are maximised but also interact with other groups to maintain a state of synergistic competition and common development. The results showed that the characteristics of ecological services values under the optimal balance state of the coastal areas in Jiangsu are compatible with the regional ecological conditions; the dominant service types (and their proportion of the total value in NRR and Non-NRR) are carbon fixation (50.70 and 50.23%), climate regulation (13.29 and 12.34%), food production (8.40 and 10.90%), water supply (8.83 and 6.60%), gas regulation (7.72 and 8.73%), and flood storage (4.95 and 5.24%). Compared to the current average service value, this status has increased by 38.61% and 23.60% respectively, in NRR and Non-NRR. The corresponding optimal population (or community) structure (and the area ratio of each population in the two regions) is dominated by forest land (40.62 and 28.74%), paddy fields (24.16 and 37.44%) and meadows (9.42 and 10.70%), and the coastal wetland populations primarily including reeds (9.78 and 3.59%), aquaculture water bodies (7.46 and 3.86%) and Spartina (3.32 and 5.60%). The structural characteristics are consistent with the development law of ecological competition and succession of the coastal wetland system in this region and the direction of ecological service value optimization. The ideas and methods proposed in this study can provide a new direction for ecological competition research.

Human disturbance is one of the essential driving forces of landscape evolution. The quantitative evaluation of the spatial and temporal characteristics of landscape evolution and its relationship with human disturbance are of great significance to regional ecological protection and management and are crucial for achieving coordinated socioeconomic development and ecological–environmental protection. In this study, we took the coastal wetlands in northern Jiangsu province, China, as the research area, and proposed a quantitative evaluation method for directional landscape evolution. On this basis, the spatiotemporal characteristics of the landscape evolution from 1980 to 2020 and the relationship with human disturbance were quantitatively evaluated by combining a human disturbance index and statistical methods. The results showed that: (1) The area of the natural wetlands decreased significantly over the past 40 years, while the areas of artificial wetlands and non-wetlands increased significantly. (2) The landscape evolution process was dominated by the degradation process. The main types of degradation were natural wetland conversion to artificial wetland and non-wetland areas and Spartina alterniflora invasion. The restoration type was mainly restoration among artificial and natural wetlands. (3) The degradation of wetland landscapes demonstrated a southward shift trend and the spatial consistency with the change of the human disturbance index was high (the correlation coefficient was 0.89). (4) The human disturbance index was significantly and positively correlated with the rate of degradation, with a correlation coefficient of 0.43, and was not significantly and positively correlated with the restoration rate, with a correlation coefficient of 0.14. The findings in this paper provide additional information and theoretical guidance for the control of coastal wetland development and utilization, as well as for achieving coordinated wetland resource development together with utilization and ecological protection in the coastal wetlands of Jiangsu province, China.

OAuth2.0 protocol has been widely adopted to simplify user login to third-party applications,at the same time,existing risk of leaking user privacy data,what even worse,causing user accounts to be hijacked.An account hijacking attack model around authorization code was built by analyzing the vulnerabilities of the OAuth2.0 protocol.A vulnerable API identification method based on differential traffic analysis and an account hijacking verification method based on authorized authentication traffic monitoring was proposed.An account hijacking attack threat detection framework OScan for OAuth2.0 authorization API was designed and implemented.Through a large-scale detection of the 3 853 authorization APIs deployed on the Alexa top 10 000 websites,360 vulnerable APIs were discovered.The further verification showed that 80 websites were found to have threat of account hijacking attack.Compared with similar tools,OScan has significant advantages in covering the number of identity provider,the number of detected relying party,as well as the integrity of risk detection.

Understanding the climatic significance of stable isotope in precipitation and ice cores on the Tibetan Plateau (TP) is of critical for the paleoclimate rebuilding. However, there is a gap between the seasonal control of precipitation and long-term isotope record from ice core. Here we present a well-dated ice core isotope record from the middle of the TP (mid-TP). Isotope variations in the past decades from this ice core show strong anti-phase relation with Southern Oscillation Index, confirming unequivocally that the large scale atmospheric circulation through the El Nino Southern Oscillation cycle, rather local climate parameters, controls the interannual signal in ice cores from the mid-TP. Results also show that the cloud top height in the northern Indian Ocean is in association with the interannual variations, confirming the same mechanism controlling the precipitation δ18O in southwest Asia and southeast Asia. The study will improve the understanding of the interannual change of Tibetan Plateau ice core isotope signal, and also the hydrological cycle in the southwest Asian region.