1. Predictive and Adaptive Bandwidth Reservation for Handoffs in QoS-Sensitive Cellular Networks IEEE Transactions on Parallel and Distributed Systems Author: Sunghyun Choi and Kang G.Shin Presented By: Yang Liu Date: Nov 26, 2002

2. Outline Introduction and Motivation System Model History-based Mobility Estimation Bandwidth Reservation and Admission Control Performance Evaluation Conclusion

3. Introduction Hand-off drops: when the cell in the new location does not have enough bandwidth to support the connection To eliminate hand-off drops Reservation of bandwidth for possible hand-offs How much bandwidth will be reserved in each cell? Per-connection bandwidth reservation By reserving each connection ’ s bandwidth in all cells the connection might pass through Not practical, and too costly if possible

4. Provided Solution Connection-level QoS parameter: - P HD : hand-off dropping probability Design goal: bounding P HD under a pre-specified target value Predictive and adaptive bandwidth reservation for hand-offs and admission control for new requests Fractional bandwidths of estimated hand-offs are reserved

5. System Model Cell indexing: (a) 1-dim. Case (b) 2-dim. Case C i,j : connection j in cell i ; b(C i,j ): its bandwidth C(i): link capacity of cell i Admission control of new connection by the BS in cell i with target reservation bandwidth B r,i :

6. Mobility Estimation Model mobile ’ s hand-off behavior and estimate mobility based on aggregate history Mobility information: when and where Observation from road traffic: O1.Traffic signals and signs(e.g., speed limits and stop signs) affect mobiles’ movements and speed significantly O2. During the rush hours, the speed of all mobiles in a given geographical area are closely correlated O3. In many cases, the direction of a mobile can be predicated from the previous path the mobile has taken so far Cell-specific history-based mobility estimation

7. Hand-off Event Quadruplets Upon the departure of a mobile from cell 0 to an adjacent cell, the BS of cell 0 caches hand-off event quadruplet (T event, prev, next, T soj ) T event : the hand-off time Prev: index of previous cell Next: index of cell after departure T soj : time duration the mobile spent at cell 0

8. Mobility Estimation Function From the cached hand-off event quadruplets(T event, prev, next, T soj ) observed during the last T int time Hand-off estimation function F HOE (t 0, prev, next, T soj ):= w n Cyclic mobility pattern w 0 = 1(today) w 1 < 1(yesterday) Describes the users ’ hand-off behaviors probabilistically given prev probability mass function for a two-dimensional random vector(next, T soj )

9. Bandwidth Reservation Mobility estimation time window: [t 0, t 0 +T est ] Hand-off probability p h (C i,j -> next): probability that C i,j hands off into cell next within time T est Extant sojourn time T ext_soj (C 0,j ): time elapsed since C 0,j entered cell 0 Example: p h (C i,j -> 4) = A/B Target reservation bandwidth at cell 0:

10. Mobility Estimation Time Window Control The larger T est, the larger p h, the larger B r,i To attain design goal more efficiently - Reference window size w (=[1/p HD,target ]) (100 if p HD,target = 0.01) - P HD < P HD,target : Should be no more than n hand-off drops out of w*n (=100*n)observed hand-offs - If violated, T est :=T est + 1 to reserve more - Otherwise, T est :=T est – 1 to reserve less Robust to (1) inaccurate mobility estimation; (2) time-varying traffic/mobility

11. Admission Control Target reservation bandwidth is calculated during the admission control phase of a new request The current cell and some adjacent cells participate in the admission control With the current target reservation bandwidth B r,i curr T1: for all, such that Calculate B r,i newly,set B r,i curr := B r,I Check if T2: Check if T3: If all the above tests are positive, then the connection is admitted.

12. Conclusion Mobility estimation based on observed history in each cell Predictive and adaptive bandwidth reservation and admission control to limit P HD below P HD,target Performance evaluation via simulation