Page Response Failure Procedures

Original Post: Page response fails, and then what?? (SIG Forum) Date: 2000-05-04

The spec states about the procedure when page response fails (pp.106) "The slave will keep listening as long as no FHS packet is received....".

There are two scenarios here --

Slave receives Master's Initial ID packet, Slave returns a ID packet, but Master doesn't receive it.
Master received the slave's ID packet, so Master returns an FHS packet, but the FHS packet doesn't get to Slave.

For the Slave, both these two scenarios are the same, i.e. The Slave has sent its response ID packet and will now keep listening as long as no Master FHS packet is received.The following show how these problems could be dealt with

Case 1

Question 1 : Case 1

So , assuming (like in case 1) the slave ID response doesn't reach the Master, so in the next slot, Master is doing page again at f(k+2) and f(k+3) but Slave is expecting FHS from Master at f(k+1). In this case, the Slave doesn't know that Master didn't receive the ID packet responding to the previous page hit and also the Slave doesn't know the Master is paging him because they are now using different frequencies.

Does this mean that the two devices will never meet until time out (pagerespTO)?, Is there any way to combat this?

Answer 1 : Case 1

Nothing per-say can be done about these "continuous" mismatches during the pagerespTO, which amounts to 8 slots (5ms) . The spec's do say that after pagerespTO has been exceeded, the slave returns back to the page scan substate for one scan period. Thus the problem could be solved that way (by simply waiting 5ms)

Question 2 : Case 1

What frequency would the slave scan at in the 'additional' scan period

Answer 2 : Case 1

Although the specs don't make it clear, one would assume it would scan at the same frequency at which the slave has successfully detected the previous page message, i.e. f'(k).

Case 2

Question 3 : Case 2

Is there any way case 2 (where Master received the slave's ID packet, so Master returns an FHS packet, but the FHS packet doesn't get to Slave) could be solved, other than waiting for pagerespTO?

Answer 3 : Case 2

Yes, according to the Baseband spec pp.136, "For each response slot the paged unit will use an X-input value one larger than in the preceding response slot.". This frequency is derived with the help of a counter value N. N is a counter starting at zero. N is increased by one when CLKE1 is set to zero as it can help resolve case 2

So in the next slot (the one after Master FHS sent at f(k+1)), Slave will send the 1st response ID packet again at f'(k+1) (increased by one when Master send out the FHS which corresponds to CLKE1 being set to zero). Master will receive this and now will send out the FHS again at f(k+2) , Slave will return 2nd response ID at f'(k+2) and so connection is now setup. So every 1.25ms, the hopping frequency is increased by one.

Note: this is fundamentally different from the first case as when a unit in the page scan substate receives an Page ID packet it freezes the 4 clock bits CLKN_16-12at their current value, thus ensuring immediate future slave hop calculations correspond to the master's estimate of the slave's hop frequency.