doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 1 NDIS Objects Byoung-Jo “J “ Kim AT&T Labs-Research Mar. ’03, Dallas

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 2 NDIS 5.1 NDIS: Device Driver to OS Interface specification for Windows (including PPC) NDIS 5.1 adds specific calls to support Win XP WLAN “auto” config – and search “802.11” –XP implements some of functions that used to be handled by vendor client software Initiate Scan and receive results Command to associate with a particular SSID Basic Configurations and Statistics

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 3

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 4 FYI: Some Interesting Behaviors If WLAN auto config profiles exist, always prefer visible SSID over invisible SSID regardless of profile order: Does not seem by design Response to Beacon WEP bit is inconsistently handled depending on card models Same for Broadcast encryption/decryption and handling of unencrypted broadcast under 1x –Hopefully will get fixed with the implementation of WPA and 11i Rumors of SSID hopping when many SSID visible –Personally encountered, but never verified in lab –Maybe fixed with SP1 or other patches

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 5 Additions for RRM Host to NIC interaction: Beyond our scope, but something to keep in mind Objects that set and retrieve TGk actions and results –Not all actions make sense to host –Hopefully, MS will to pick a subset of what we decide as standards Extend the statistics Object and Fix definitions –Hopefully, it will share our definitions and parameters Can be used for SNMP if an agent is present

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 6 Linux Wireless Tools Linux/Tools.htmlhttp:// Linux/Tools.html Adds WLAN specific APIs for drivers and some tools –Available in most Linux distributions now –Open source, voluntary effort still progressing Drivers implementing the API available for several major chipsets and NIC models Relaxed approach like many open source projects

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 7 Some interesting Parmeters #define SIOCSIWSENS 0x8B08 /* set sensitivity (dBm) */ #define SIOCSIWTXPOW 0x8B26 /* set transmit power (dBm) */ #define IWEVQUAL 0x8C01 /* Quality part of statistics (scan) */ #define IWEVREGISTERED 0x8C03 /* Discovered a new node (AP mode) */ #define IWEVEXPIRED 0x8C04 /* Expired a node (AP mode) */ #define IW_SCAN_ALL_ESSID 0x0001 /* Scan all ESSIDs */ #define IW_SCAN_ALL_FREQ 0x0004 /* Scan all Frequencies */ #define IW_SCAN_ALL_MODE 0x0010 /* Scan all Modes */ #define IW_SCAN_ALL_RATE 0x0040 /* Scan all Bit-Rates */ struct iw_quality qual; /* Quality of the link * (instant/mean/max) */ struct iw_discarded discard; /* Packet discarded counts */ struct iw_missed miss; /* Packet missed counts */

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 8 Continued struct iw_range { /* Informative stuff (to choose between different interface) */ __u32 throughput; /* To give an idea... */ /* In theory this value should be the maximum benchmarked * TCP/IP throughput, because with most of these devices the * bit rate is meaningless (overhead an co) to estimate how * fast the connection will go and pick the fastest one. * I suggest people to play with Netperf or any benchmark... */ struct iw_quality max_qual; /* Quality of the link */

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 9 Continued /* Average quality of link & SNR */ struct iw_quality avg_qual; /* Quality of the link */ /* This should contain the average/typical values of the quality * indicator. This should be the threshold between a "good" and * a "bad" link (example : monitor going from green to orange). * Currently, user space apps like quality monitors don't have any * way to calibrate the measurement. With this, they can split * the range between 0 and max_qual in different quality level * (using a geometric subdivision centered on the average). * I expect that people doing the user space apps will feedback * us on which value we need to put in each driver... */

doc.: IEEE /172r0 Submission March 2003 Byoung-Jo Kim, AT&TSlide 10 Recommendations Though beyond our scope, we can interact informally for feedback and feedforward Jointly develop optional recommendation for accessing TGk features within host