SensorNetGuard: A Dataset for Identifying Malicious Sensor Nodes

The dataset, titled "SensorNetGuard: A Dataset for Identifying Malicious Sensor Nodes," comprises 10,000 samples with 21 features. It is designed to facilitate the identification of malicious sensor nodes in a network environment, specifically focusing on IoT-based sensor networks.General Metrics§ Node ID: The unique identifier for each node.§ Timestamp: The time at which data or a packet is sent or received.§ IP Address: Internet Protocol address of the node.Network Traffic Metrics§ Packet Rate: The number of packets sent/received per unit time.§ Packet Drop Rate: The rate at which packets are being dropped.§ Packet Duplication Rate: The rate at which packets are duplicated.§ Data Throughput: The amount of data successfully transferred from one point to another in a given time frame.Signal Metrics§ Signal Strength: Measured in dBm, indicates the power level of the signal.§ Signal-to-Noise Ratio (SNR): Measures the clarity of the signal.Power Usage Metrics§ Battery Level: Remaining power in the node.§ Energy Consumption Rate: The rate at which the node consumes energy.Routing Metrics§ Number of Neighbors: The count of nodes within direct communication range.§ Route Request Frequency: The frequency with which a node is asking for routes.§ Route Reply Frequency: The frequency with which a node responds to route requests.Behavioral Metrics§ Data Transmission Frequency: How often a node sends data.§ Data Reception Frequency: How often a node receives data.§ Error Rate: The number of erroneous packets per unit time.Miscellaneous Metrics§ CPU Usage: The percentage of CPU being used by the node.§ Memory Usage: The amount of RAM being used.§ Bandwidth: Available data transmission rate.Metrics Specific to Attacks§ Is_Malicious: A binary flag (0 or 1) indicating whether the node is malicious.The dataset includes a diverse range of features that allow for the application of machine learning models to identify various types of attacks, such as black hole, gray hole, flooding attacks, and Sybil attacks. Some features may be more relevant for specific types of attacks, so feature engineering and selection are crucial steps. Various machine learning algorithms can be trained on this dataset. The "Is_Malicious" column serves as the ground truth for learning algorithm. For unsupervised learning, anomaly detection algorithms can be used to identify outliers or anomalies that may correspond to malicious activities.Being able to accurately identify malicious nodes can significantly improve the security posture of sensor networks, particularly in IoT environments where such attacks can have severe consequences.By training a model on this dataset, network administrators can identify potential security threats in real-time, thus enabling proactive measures to isolate or remove malicious nodes.Ultimately, the dataset serves as a comprehensive resource for developing and validating machine learning models aimed at enhancing network security by identifying malicious sensor nodes.

该数据集名为“SensorNetGuard：用于识别恶意传感器节点的数据集”，包含10,000个样本，具有21个特征。该数据集旨在促进网络环境中恶意传感器节点的识别，特别是针对基于物联网的传感器网络。在一般指标方面，包括节点ID（每个节点的唯一标识符）、时间戳（数据或数据包发送或接收的时间）、IP地址（节点的互联网协议地址）。在网络流量指标方面，包含数据包速率（每单位时间内发送/接收的数据包数量）、数据包丢失速率（数据包丢失的速率）、数据包重复速率（数据包重复的速率）和数据吞吐量（在给定时间框架内从一点到另一点成功传输的数据量）。在信号指标方面，包括信号强度（以dBm为单位，表示信号的功率水平）和信噪比（SNR，衡量信号的清晰度）。在功耗指标方面，包括电池电量（节点剩余的电量）和能耗速率（节点消耗能量的速率）。在路由指标方面，包括邻居数量（直接通信范围内的节点数量）、路由请求频率（节点请求路由的频率）和路由回复频率（节点响应路由请求的频率）。在行为指标方面，包括数据传输频率（节点发送数据的频率）、数据接收频率（节点接收数据的频率）和错误率（每单位时间内错误数据包的数量）。在杂项指标方面，包括CPU使用率（节点使用的CPU百分比）、内存使用量（使用的RAM量）和带宽（可用的数据传输速率）。针对攻击的特定指标包括是否恶意（一个二进制标志，表示节点是否恶意）。该数据集包含多种多样的特征，允许应用机器学习模型以识别各种类型的攻击，如黑洞攻击、灰色洞攻击、洪水攻击和Sybil攻击。某些特征可能对特定类型的攻击更为相关，因此特征工程和选择是至关重要的步骤。多种机器学习算法可以在此数据集上训练。'Is_Malicious'列作为学习算法的基准。对于无监督学习，可以使用异常检测算法来识别可能与恶意活动相关的异常或离群值。能够准确识别恶意节点可以显著提升传感器网络的安全态势，特别是在物联网环境中，此类攻击可能带来严重后果。通过在此数据集上训练模型，网络管理员可以实时识别潜在的安全威胁，从而采取主动措施隔离或移除恶意节点。最终，该数据集为开发和验证旨在通过识别恶意传感器节点来增强网络安全的机器学习模型提供了全面的资源。