A Dataset for Exploring Wi-Fi Network Diversity in Vehicle-to-Infrastructure Communication

Introduction: This dataset contains space and time-indexed performance data for Wi-Fi communication between a moving vehicle and a set of stationary Access Points (APs). In order to allow comparisons between technologies, 3 different types of Wi-Fi are used in parallel: 800.11n, ac, and ad. For more information, please consult the following article: Exploring Wi-Fi Network Diversity for Vehicle-to-Infrastructure Communication, Rui Meireles, António Rodrigues, Andrei Stanciu, Ana Aguiar, Peter Steenkiste, in the 2020 IEEE Vehicular Networking Conference (VNC 2020), December 2020, doi:10.1109/VNC51378.2020.9318407. Video presentation available here. Experiment description: The AP was placed at the corner of a residential area intersection while the mobile client drove a circuit around it. The mobility pattern is shown in the animated file vehicle-movement.gif. The data is divided into traces, gathered on different dates, using different vehicles to support the AP on the roof, as shown below: trace nr date start time n(n) n(ac) n(ad) AP vehicle n(clients) 302 2019-08-20 10:28:45 3262 2787 423 2001 Honda Civic sedan 1 303 2019-08-20 11:26:23 3374 3027 312 - 2 304 2019-08-20 12:39:46 1711 216 14 - 3 (n & ac) 2 (ad) 401 2019-08-22 10:19:24 1685 1681 545 2003 Peugeot Partner 1 402 2019-08-22 10:48:26 2859 2827 764 - 2 403 2019-08-22 11:39:36 135 135 116 - 2 404 2019-08-22 11:42:50 114 114 53 - 2 405 2019-08-22 11:45:07 2019 2019 507 - 2 APs: all positioned at coordinates {lat : 41.111879, lon : -8.631146} ap device 802.11 type channel cntr. freq (MHz) bw (MHz) unifi-003 ubiquiti ac lite n 6 2437 20 unifi-001 - ac 40 5200 40 tp-01 tp-link talon ad7200* ad 1 60480 2160 *running tp-link's original firmware, not OpenWrt Main clients: all positioned in the moving vehicle's roof, a vw golf mk3 802.11 type radio nr. antennas laptop n csl usb 2.0 wlan Adapter 300 Mbps 2 m1 ac tp-link archer t4uh 2 w4 ad tp-link talon ad7200 (tp-03) - w4 Background clients: the purpose is to increase channel util. 802.11 type radio nr. antennas laptop position n tp-link wn722n 1 w2 fixed, ~2m away from AP n tp-link wn722n 1 w3 '' ac csl usb 2.0 wlan Adapter 300 Mbps 2 w2 '' ac csl usb 2.0 wlan Adapter 300 Mbps 2 w3 '' ad tp-link talon ad7200 (tp-04) - macbook stopped vehicle's roof Monitor nodes: all positioned in the moving vehicle's roof 802.11 type radio nr. antennas laptop n csl usb 2.0 wlan Adapter 300 Mbps 2 m1 ac tp-link talon ad7200 (tp-02) 8 w4 ad tp-link talon ad7200 (tp-02) - w4 Dataset structure: All the packet captures are already digested and ready to use in the file wifi-exp-log-summary.csv. An explanation of the fields below: systime : system time (1 Hz resolution) that this row refers to. All node clocks were synchronized through NTP. traceNr : nr. of the trace the row belongs to. lon : longitude (in degrees) reported by the receiver's GPS at systime lat : latitude reported by the receiver's GPS at systime receiverAlt : altitude (in meters) reported by the receiver's GPS at systime receiverX : x coordinate of the receiver's position when space is discretized as a Cartesian plane and the sender is set to be the origin of the coordinate system. The x axis corresponds to east-west (positive values are east, negative values are west). Unit is meters. receiverY : y coordinate of the receiver's position when space is discretized as a Cartesian plane receiverDist : distance (in meters) of receiver to ap(s) receiverSpeed : speed (in m/s) reported by the receiver's GPS at systime receiverId : system-specific id for the client (in the vehicle) senderId : system-specific id for the ap serving the client (side of the road) isIperfOn : 1 if row's systime corresponds to a period where our UDP packet consumer application is known to have been running on the receiver side. isInLap : 1 if this row's systime has been marked as being part of a time period where clients were doing laps around the APs, 0 otherwise. rssiMean : the mean of the RSSI (Received Signal Strength Indicator) values of frames received by the client from the ap during the 1-second period systime period the row refers to. snrMean : SNR (signal to noise ratio) retrieved from 802.11ad sectore sweep frames. channelFreq : center frequency of the WiFi channel used, in MHz. channelBw : bandwidth of the WiFi channel used, in MHz. channelUtil : percentage of time the wireless medium was sensed to be busy during the 1-second period systime period the row refers to. In traces 40x, the 802.11n and ac routers didn't log channel busy time, and as such we had to approximate channel util. based on x,y coordinates and nr. of active clients. wifiType : 802.11 type (e.g., n, ac or ad). nrClients : nr. of parallel clients operating in wifiType mode, on the same channel and bandwidth as receiverId. dataRateMedian : the median of the bitrate values of frames received by the client from the ap during the 1-second period systime period the row refers to. dataRateMean : the mean of the bitrate values of frames received by the client from the ap during the 1-second period systime period the row refers to. nBytesReceived : total number of bytes received by the client from the ap during the 1-second period systime period the row refers to. tghptConsumer : throughput reported by the UDP packet consumer application, during the 1-second period systime period the row refers to. nRetries : nr. of WLAN-level re-transmissions on 1 second period meanBeaconRssi : mean RSSI measured from beacons in 1 second period. nan values are filled with -100 dBm. meanInterBeaconTime : mean interval between consecutive beacons, within 1 second period. nan values are filled with 1 sec. nBeacons : total nr. of beacons received by client within 1 second period.

## 数据集说明

### 引言
本数据集包含移动车辆与一组固定接入点(Access Points, APs)之间的Wi-Fi通信性能时序时空索引数据。为支持不同技术间的性能对比，本次实验并行采用了3种Wi-Fi标准：802.11n、ac及ad。如需了解更多细节，请参阅以下论文：*Exploring Wi-Fi Network Diversity for Vehicle-to-Infrastructure Communication*，作者为Rui Meireles、António Rodrigues、Andrei Stanciu、Ana Aguiar、Peter Steenkiste，发表于2020年IEEE车载网络会议(IEEE Vehicular Networking Conference, VNC 2020)，2020年12月，DOI: 10.1109/VNC51378.2020.9318407。相关视频演示可在此处获取。

### 实验设置
接入点部署于居民区交叉口拐角处，移动客户端沿环绕该拐角的环路行驶，其移动轨迹可查看动画文件`vehicle-movement.gif`。本次实验按不同日期、使用不同搭载接入点的车辆采集多组轨迹数据，具体参数如下：

| 轨迹编号 | 日期 | 开始时间 | 802.11n采样数 | 802.11ac采样数 | 802.11ad采样数 | 接入点型号 | 测试车辆 | 客户端数量 |
|----------|------------|------------------|--------------|---------------|---------------|------------------------|------------------------|------------|
| 302 | 2019-08-20 | 10:28:45 | 3262 | 2787 | 423 | 2001 | 本田思域(Honda Civic)轿车 | 1 |
| 303 | 2019-08-20 | 11:26:23 | 3374 | 3027 | 312 | - | - | 2 |
| 304 | 2019-08-20 | 12:39:46 | 1711 | 216 | 14 | 3（仅支持802.11n与ac） | 2（仅支持802.11ad） | 4 |
| 401 | 2019-08-22 | 10:19:24 | 1685 | 1681 | 545 | 2003 | 标致伙伴(Peugeot Partner) | 1 |
| 402 | 2019-08-22 | 10:48:26 | 2859 | 2827 | 764 | - | - | 2 |
| 403 | 2019-08-22 | 11:39:36 | 135 | 135 | 116 | - | - | 2 |
| 404 | 2019-08-22 | 11:42:50 | 114 | 114 | 53 | - | - | 2 |
| 405 | 2019-08-22 | 11:45:07 | 2019 | 2019 | 507 | - | - | 2 |

### 接入点配置
所有接入点均部署于坐标`{纬度: 41.111879, 经度: -8.631146}`处，各接入点的详细参数如下：

| AP名称 | 设备型号 | 802.11标准 | 信道 | 中心频率(MHz) | 带宽(MHz) |
|------------|-------------------------|------------|------|---------------|-----------|
| unifi-003 | 优倍快(ubiquiti) ac lite | n | 6 | 2437 | 20 |
| unifi-001 | - | ac | 40 | 5200 | 40 |
| tp-01 | TP-Link Talon AD7200* | ad | 1 | 60480 | 2160 |

*注：tp-01运行TP-Link原厂固件，未刷写OpenWrt系统。

### 主客户端配置
所有主客户端均部署于移动车辆车顶，车型为大众高尔夫MK3(Volkswagen Golf MK3)，各主客户端的参数如下：

| 802.11标准 | 无线网卡型号 | 天线数量 | 标识 |
|------------|---------------------------------------|----------|-------|
| n | CSL USB 2.0 WLAN Adapter 300 Mbps | 2 | m1 |
| ac | TP-Link Archer T4UH | 2 | w4 |
| ad | TP-Link Talon AD7200 (tp-03) | - | w4 |

### 背景客户端配置
背景客户端用于增加信道利用率，其部署参数如下：

| 802.11标准 | 无线网卡型号 | 天线数量 | 标识 | 部署位置 |
|------------|---------------------------------------|----------|------|------------------------------|
| n | TP-Link WN722N | 1 | w2 | 固定部署，距接入点约2米处 |
| n | TP-Link WN722N | 1 | w3 | 固定部署，距接入点约2米处 |
| ac | CSL USB 2.0 WLAN Adapter 300 Mbps | 2 | w2 | 固定部署，距接入点约2米处 |
| ac | CSL USB 2.0 WLAN Adapter 300 Mbps | 2 | w3 | 固定部署，距接入点约2米处 |
| ad | TP-Link Talon AD7200 (tp-04) | - | - | 固定部署，距接入点约2米处 |

### 监控节点配置
所有监控节点均部署于移动车辆车顶，各节点参数如下：

| 802.11标准 | 无线网卡型号 | 天线数量 | 标识 |
|------------|---------------------------------------|----------|-------|
| n | CSL USB 2.0 WLAN Adapter 300 Mbps | 2 | m1 |
| ac | TP-Link Talon AD7200 (tp-02) | 8 | w4 |
| ad | TP-Link Talon AD7200 (tp-02) | - | w4 |

### 数据集结构
本数据集的所有报文捕获结果均已预处理完成，可直接使用的汇总数据文件为`wifi-exp-log-summary.csv`，各字段说明如下：
1. `systime`：本条数据对应的系统时间，分辨率为1Hz。所有节点时钟均通过网络时间协议(Network Time Protocol, NTP)同步。
2. `traceNr`：本条数据所属的轨迹编号。
3. `lon`：本条数据采集时刻，接收端GPS报告的经度（单位：度）。
4. `lat`：本条数据采集时刻，接收端GPS报告的纬度（单位：度）。
5. `receiverAlt`：本条数据采集时刻，接收端GPS报告的海拔高度（单位：米）。
6. `receiverX`：将空间离散为笛卡尔坐标系，并以接入点为坐标原点时，接收端位置的X坐标。X轴对应东西方向（正值为东，负值为西），单位为米。
7. `receiverY`：上述笛卡尔坐标系下，接收端位置的Y坐标，单位为米。
8. `receiverDist`：接收端与接入点之间的距离（单位：米）。
9. `receiverSpeed`：本条数据采集时刻，接收端GPS报告的行驶速度（单位：m/s）。
10. `receiverId`：车辆内客户端的系统专属标识。
11. `senderId`：为当前客户端提供服务的接入点的系统专属标识（对应道路侧）。
12. `isIperfOn`：标记位，若本条数据对应的系统时间内，接收端运行了UDP报文消费应用，则取值为1，否则为0。
13. `isInLap`：标记位，若本条数据对应的系统时间属于客户端环绕接入点行驶的周期，则取值为1，否则为0。
14. `rssiMean`：本条数据对应1秒周期内，客户端从接入点接收的帧的接收信号强度指示(Received Signal Strength Indicator, RSSI)平均值。
15. `snrMean`：从802.11ad扇区扫描帧中提取的信噪比(Signal to Noise Ratio, SNR)平均值。
16. `channelFreq`：当前使用的Wi-Fi信道的中心频率，单位为MHz。
17. `channelBw`：当前使用的Wi-Fi信道的带宽，单位为MHz。
18. `channelUtil`：本条数据对应1秒周期内，无线介质被检测为繁忙的时间占比。在轨迹40x系列中，802.11n与ac路由器未记录信道繁忙时间，因此需基于X、Y坐标与活跃客户端数量估算信道利用率。
19. `wifiType`：当前使用的802.11标准类型（如n、ac或ad）。
20. `nrClients`：当前信道与带宽下，以相同`receiverId`运行的同类型并行客户端数量。
21. `dataRateMedian`：本条数据对应1秒周期内，客户端从接入点接收的帧的比特率中位数。
22. `dataRateMean`：本条数据对应1秒周期内，客户端从接入点接收的帧的比特率平均值。
23. `nBytesReceived`：本条数据对应1秒周期内，客户端从接入点接收的总字节数。
24. `tghptConsumer`：本条数据对应1秒周期内，UDP报文消费应用报告的吞吐量。
25. `nRetries`：本条数据对应1秒周期内，WLAN层的重传次数。
26. `meanBeaconRssi`：本条数据对应1秒周期内，从接入点信标帧中测得的平均RSSI。缺失值填充为-100 dBm。
27. `meanInterBeaconTime`：本条数据对应1秒周期内，相邻信标帧之间的平均间隔。缺失值填充为1秒。
28. `nBeacons`：本条数据对应1秒周期内，客户端接收的信标帧总数量。