Frame Level Troubleshooting
What is a Frame
PMP 450/450i product series uses a TDD (Time Division Duplexing) frame structure in which the Access Point (AP) and the Subscriber Module (SM) transmit and receive in non-overlapping times using the same frequency. Each transmission frame is divided into two portions: a downlink and an uplink. There are scheduled and unscheduled slots within the downlink and uplink slots. The below image shows the high level components of the frame.
Downlink Slots
During the downlink time of the TDD frame, the AP transmits and the SM receives. The unscheduled downlink slots primarily consist of preamble, beacon, uplink scheduling map, BER symbol. These facilitates time/frequency synchronization, carry radio configuration information, etc. The scheduled downlink slots carry the control and data messages. We can classify the scheduled downlink slots into the following categories:
- Registration and Control Information
This consist of the critical control information and has the highest priority. They need to be received as fast as possible - High Priority Data
This consist of all high priority data which is typically used to support low latency traffic such as VoIP, video streaming etc. The TOS field of the IPv4 header of the data traffic is one method used to classify data into high priority data. The downlink scheduler makes sure enough priority is given for scheduling this data for transmission. - Low Priority Data
This consist of the regular data traffic. Most of the data sent across to the SMs fall into this category. - Broadcast or Multicast Data
All broadcast and multicast packets are scheduled in this category. - Acknowledgement
Acknowledgments for data transmitted in the uplink are sent in the downlink of the next frame
Uplink Slots
During the uplink time of the TDD frame, the SM transmits and the AP receives. The unscheduled uplink slots are used for contention. Any scheduled portion of the uplink which is under-utilized, is also used for contention. The contention slots could be used for network entry or bandwidth request by every SM. The scheduled uplink slots carry the control and data messages. We can classify the scheduled uplink slots into the following categories:
- High Priority Data
This is similar to the downlink high priority data in the opposite direction. - Low Priority Data
This consist of the regular data traffic similar to the downlink low priority data. - Acknowledgement
Acknowledgments for data transmitted in the downlink are sent in the uplink of the next frame
Frame Configuration
The below figure taken from Configuration->Radio of the web GUI shows the specific values which can be configured for a frame. Please refer to the PMP User guide to know more information on these configuration.
Further the frame time depends on the frame period configured which can be either 5 or 2.5 milliseconds. The downlink and uplink times are divided into OFDM symbols; the symbol time depends on the channel bandwidth, and the cyclic prefix length.
Frame Utilization Statistics
Efficient use of the TDD frame will help optimize the overall customer experience. The frame utilization statistics is the diagnostic tool which gives detailed information on the real time usage of the TDD frame, giving an understanding of how effectively the RF channel is utilized. This tool is available only on the AP and can be found in Statistics->Frame Utilization page of the web GUI. The feature was initially added into the 13.4 release and was enhanced in the subsequent releases.
Diagnostic Interval
Frame utilization can be diagnosed in 3 different intervals which are 1, 5 and 15 minutes respectively. The below image shows the section which selects which interval is to be used. The time required for the next update is also shown for convenience so that the user is aware of how long he needs to wait for the next set of statistics.
Frame Utilization Summary
The below image shows the section which displays the summary of the overall frame utilization for the given interval. This is displayed in percentage. Customers will most likely need a threshold that is acceptable. One customer may consider an overall downlink or uplink frame utilization of 75% as acceptable, while another may be happy with an overall frame utilization of 90%.
Detailed downlink utilization
The below image shows the section which displays the detailed counters of downlink slots utilized in the last interval. The statistics displayed are only for the scheduled portion of the frame. The total number of slots used and a per-frame average is provided for convenience. The per-frame average is computed using the total downlink slots used and the total number of frames possible for the given diagnostic interval. The Low Priority, High Priority, Broadcast/Multicast, Acknowledgements are the detailed categorized counters which can give an idea of what type of data or control traffic were the frames consumed for.
Detailed uplink utilization
The below image shows the section which displays the detailed counters of downlink slots utilized in the last interval. The total number of slots and per-frame average is provided similar to the downlink section. The Low Priority, High Priority and Acknowledgements provide the detailed categorized counters for specific troubleshooting. The Contention slots actually used is provided and a per-frame average is also displayed for convenience. The number of bandwidth requests received and missed is also displayed.
Maximum Possible Counts
The maximum possible downlink, uplink and contention slot counts are displayed as shown in the below figure. These counts are based on the user configured values of frame configuration as mentioned in section ‘Frame Configuration’ above. These counts are also used to compute the overall frame utilization summary.
Packet discard
Every user would like to utilize the channel as much as possible and would like to reach a frame utilization close to 100%. However, over utilization of the channel needs to be detected to understand that the frame utilization is exceeded its maximum and some action needs to be taken. The below image shows the section which shows Ethernet and Radio packet discard counts. Once these values start increasing considerably, the user may have to re-check the frame configuration or the radio installation.
cnMedusa Frame Utilization
The PMP 450m with MU-MIMO technology is designed to operate with higher frame utilization. When operating in MU-MIMO mode, each downlink slot in the frame may be used to send data to up to seven SMs. For non MU-MIMO platforms, the frame utilization may reach 100% with each slot carrying data to a single SM. With MU-MIMO, this does not indicate that the downlink is at maximum capacity. Other, suitably positioned SMs may send data in parallel in this case. Hence even at 100% frame utilization, there may be much more capacity in the downlink and the frame utilization displayed similar to the non MU-MIMO platforms could be misleading.
Backhaul Frame Utilization
In case of backhaul, both master and slave show frame utilization statistics. However, the Backhaul master will have only the downlink statistics and the Backhaul slave will have only the uplink statistics. This is because the scheduler runs only in one direction in case of backhaul.
Fetching via SNMP
All these Frame utilization values displayed in the web GUI are also available via SNMP. Statistics at all 3 intervals are available. Users can programmatically analyse the statistics by capturing the data at the required interval.
The MIB file WHISP-APS-MIB.txt contains the OID information. The below MIB tree shows the groups of OIDs available for frame utilization.
The parent OID group for frame utilization statistics is whispApsFrUtlStats (.1.3.6.1.4.1.161.19.3.1.12). The OIDs are segregated into 3 interval groups, namely whispApsFrUtlStatsIntervalLow (.1.3.6.1.4.1.161.19.3.1.12.1), whispApsFrUtlStatsIntervalMedium (.1.3.6.1.4.1.161.19.3.1.12.2) and whispApsFrUtlStatsIntervalHigh (.1.3.6.1.4.1.161.19.3.1.12.3) which correspond to 1, 5 and 15 minute intervals respectively. An OID group named whisAPsFrUtlStatsInstantaneous (.1.3.6.1.4.1.161.19.3.1.12.4) is provided for instantaneous frame utilization. This is applicable only for the cnMedusa product line as of now.
The below snip of the MIB tree shows the OIDs for the 1 minute (low) interval statistics. Similar set of OIDs are available for 5 minute (medium) and 15 minute (high) intervals.
The OID names are self-explanatory and correspond to the web GUI entries. The MIB description for each SNMP OIDs provides the detailed information on each OID.
Remote analysis using cnMaestro
In case the devices are on-boarded to cnMaestro, the frame utilization statistics can be viewed using the cnMaestro web interface.
The dashboard has the frame utilization summary for uplink and downlink displayed on the top right corner for the PMP AP device. The below figure shows this.
The same values can also be seen in the statistics table as shown in the image below. After selecting the desired sector on cnMaestro, select PMP AP as device type and click on the Statistics grid. The frame utilization values can be seen in the traffic section of statistics.
The below image shows the performance graph which is also plotted with respect to the uplink and downlink frame utilization values. Select the desired AP among the sectors in cnMaestro and click on the Performance tab. The user has options to switch between the aggregate of 4 hours to 1 week. The statistics data which is used to plot the graph is retrieved from the device at 5 minute intervals.
Remedial actions
When a poor frame utilization is observed, some of the below remedial actions could be employed to improve the utilization.
- Further sectorization – Pulling in more AP/SM devices and breaking down into more sectors. This may involve site planning, re-configuring APs in the sectors, etc.
- Downlink/uplink ratio – The frame configuration on the AP can be tuned to increase or decrease the downlink percentage, based on the nature of traffic in the sector.
- Contention slots – The frame configuration on the AP can be tuned to have more or less contention slots.
- Improved RF links – This involves optimizing the RF link ensuring everything has been done to obtain the highest possible modulation for the link.
- Reshaping user data profiles – Proper MIR can be employed to limit the amount of data in case of extreme data usage. Review the CIR and make sure if its utilized effectively.