A Pseudo Random Coordinated Scheduling Algorithm for Bluetooth Scatternets MobiHoc 2001

Outline  Abstract  Introduction  Overview of the PCSS algorithm  Operation of PCSS  Reference algorithm  Simulation results  Conclusions

Abstract  To schedule communication with bridging nodes one must take into account their availability in the different piconets.  This algorithm assign meeting points with their peers such that the sequence of meeting points.

Introduction  Two important phenomena that can reduce the efficiency of the polling based communication. Slaves that have no data to transmit. At the time of an expected poll one of the nodes of a master-slave node pair may not be present in the piconet.

Introduction  Hard coordination Eliminates ambiguity with regards to a node ’ s presence in piconets, but it implies a complex, scatternet wide coordination problem.

Introduction  Soft coordination Nodes decide their presence in piconets based on local information.

Introduction  Two key component of this algorithm Checkpoints –Serve as regular meeting points for neighboring nodes when they can exchange packets. Dynamic adjustment of checking intensity –Bandwidth can be allocated and deallocated to a particular link.

Overview of the PCSS algorithm  A node remains active on the current link until there is user data in both directions or until it has to leave for a next checkpoint.  A pseudo random sequence that is generated based on Bluetooth clock of the master MAC address of the slave

Overview of the PCSS algorithm Current base checking interval

Operation of PCSS  Initialization  Communication  Increasing and Decreasing Checking Intensity

Initialization  A master slave node pair share the same master ’ s clock and slave ’ s MAC address information Guaranteed that the same pseudo random sequence will be produced at each node.

Communication  A node remains active on the current link until there is user data in both directions or until it has to leave for a next checkpoint.  During the communication any of the nodes can leave in order to attend a coming checkpoint on one of its other links.

Communication  After passing a checkpoint the variable t (i) check is updated to the next checkpoint by pseudo random generator(PseudoChkGen). t (i) check =PseudoChkGen(T (i) check, A (i) slave,t (i) )  There is a maximum and minimum checking interval T max =2 f max and 2 f min.

Increasing & decreasing checking intensity  To measure the utilization of checkpoints  (i) on t the i th link of the node.  If the checkpoint has been utilized the variable  (i) is updated as  (i) =q uti *  (i) + (1-q uti ) * 1 0<=q uti <1  If the checkpoint has not been utilized it is updated as  (i) =q uti *  (i) + (1-q uti ) * 0 0<=q uti <1

Increasing & decreasing checking intensity  Each node measures its own utilization  (node) and updates the  (node) variable after each N uti,win number of slots as fllows  (node) =q uti (node) *  (node) + (1-q uti (node) ) *  (win) – where  (win) is the fraction of time slots in the past time window of length N uti,win where the node has been active over the total number of time slots N uti,win.

Increasing & decreasing checking intensity  If the utilization of checkpoints on link i falls below the lower threshold  lower, the current base checking period T (i) check will be doubled.

Increasing & decreasing checking intensity

 A checking intensity increase is performed on link I if the following two conditions are satisfied  (i) >  upper and  (node) <  (node) upper

Increasing & decreasing checking intensity

Reference algorithms  Ideal Coordinated Scatternet Scheduler(ICSS). A node is aware of the already pre-scheduled transmissions of its neighbors. A node is aware of the content of the transmission buffers of its neighbors.  Uncoordinated Greedy Scatternet Scheduler(UGSS). Each node visits its neighbors in a random order.

Reference algorithms

Simulation results  Network Access Point Scenario.

Simulation results

 Impact of number of forwarding hops.

Simulation results

 Impact of bridging degree.

Simulation results

Conclusions  This algorithm that can efficiently control communication in Bluetooth scatternets with out exchange of control information between Bluetooth device.  Two key components Pseudo random sequences of meeting points. A set of rules that govern the increase and decrease of meeting point intensity.