130 likes | 144 Views
This article discusses the problem of detecting periodic radars using periodic quiet times and proposes remedies such as randomization of Quiet Offset and introducing appropriate redundancy. The advantages of these solutions are statistically guaranteed detection, robustness, sleep mode compatibility, and operation in both BSS and IBSS.
E N D
Detection of Periodic Radars Peter Larsson Ericsson Research peter.larsson@era.ericsson.se Peter Larsson, Ericsson Research
Problem • Detection of periodic radars by means of periodic quiet times is ”unsuitable”. Peter Larsson, Ericsson Research
Radar pulse Periodic Radars &Periodic Quiet Times • Example TQuiet_Duration Quiet time Quiet time Quiet time Quiet time Beacon Beacon Beacon Beacon Beacon Time TQuiet_Offset TQuiet_Offset TQuiet_Offset TQuiet_Offset • Does this occur in reality? • The draft states”All STAs in an IBSS shall continue to schedule the quiet periods by including a Quiet element in any Beacons or Probe Responses, with the same Quiet Period, Quiet Duration and Quiet Offset, if any previous Beacons or Probe Responses in the IBSS contained a Quiet element” Peter Larsson, Ericsson Research
Periodic Radars & Periodic Quiet Times • Zero probability of detecting periodic radars when n/PRF=Quiet_period_field·Beacon_interval, where n is an integer. • Long detection time of periodic radars, when Quiet times and Radar Pulses are (slowly) sliding relative to each other. • It is a real problem in an IBSS. • It is also a real problem in a BSS when fixed quiet parameters are deployed. Peter Larsson, Ericsson Research
Radar pulse Quiet time Quiet time Quiet time Quiet time Beacon Beacon Beacon Beacon Beacon Time TQuiet_Offset TQuiet_Offset TQuiet_Offset TQuiet_Offset The remedy - part 1 • Randomization of the Quiet Offset! Peter Larsson, Ericsson Research
STA may miss Beacon and then accidently TX during the Quiet time STA TXing Sleep time Randomization partly enabled by current BSS Solution • The AP may change the value of the Quiet Offset field, provided scheduling of Quiet times is stopped first. Quiet time Quiet time Quiet time Quiet time Beacon Beacon Beacon Beacon Beacon Time Quiet Offset1 Quiet Offset2 Quiet Offset3 Peter Larsson, Ericsson Research
Problems with current BSS solution • The ”existing” randomization operation: • Only applies to BSS, but not to IBSS • Does not provide desired robustness • Is not ”sleep mode compatible” Peter Larsson, Ericsson Research
The remedy - part 2 • Introduce appropriate redundancy. Peter Larsson, Ericsson Research
Quiet Offsetn+1 Quiet Offsetn+2 Quiet Offsetn+N Quiet Offsetn+2 Quiet Offsetn+3 Quiet Offsetn+1+N Main Solution • Multiple Quiet Offsets indicated in each Beacon Quiet time Quiet time Quiet time Quiet time Quiet time Beacon Beacon Beacon Beacon Beacon Time TQuiet_Offset TQuiet_Offset TQuiet_Offset TQuiet_Offset TQuiet_Offset Peter Larsson, Ericsson Research
Quiet Period Element ID (40) Quiet Duration Length (6) Quiet Count 1 1 1 2 1 New Quiet Element Format Octets: Quiet Offset n Quiet Offset n+1 Quiet Offset n+2 2 2 2 Octets: Peter Larsson, Ericsson Research
Alternative Solution • Indicate a state for a pseudo-random generator in the Beacon • Derive the Quiet Offset from the state • Update the state for every Beacon DelayedBeacon Quiet time Quiet time Beacon Beacon Time TQuiet_Offset TQuiet_Duration TQuiet_Offset TQuiet_Duration TBTT TBTT TBeacon_Interval Quiet Offset state N.B. Beacon interval used for illustration here. DCF Interval may be used instead Peter Larsson, Ericsson Research
Element ID (40) Length (6) Quiet Count Quiet Period Quiet Duration 1 1 1 1 2 New Quiet Element Format Octets: Quiet Offset state 2 Octets: Peter Larsson, Ericsson Research
Conclusions • Current Quiet time method unsuitable for detection of periodic radars • Introduce quiet time randomization and appropriate redundancy • Advantages • Statistically guaranteed detection* • Robustness • Sleep mode compatibility • Operation in both BSS and IBSS • The Main Solution is more flexible than the alternative solution, but the latter is more compact and suits IBSS well *N.B. Not entirely true for the IBSS case using the main solution, but yet sufficiently good Peter Larsson, Ericsson Research