1. Ghost Femtocells: a Novel Radio Resource Management Scheme for OFDMA Based Networks WCNC 2011

2. Outline  Introduction  System Model  Ghost Femtocells: The Proposed Resource Allocation Algorithm  Simulation  Conclusions 2

3. Introduction  The femtocell deployment in 3GPP/LTE sets new challenges to interference mitigation techniques and Radio Resource Management (RRM).  The goal of this paper is to achieve effective spectral reuse between macrocells and femtocells while guaranteeing the QoS of users served by both macro and femto base stations. 3

4. System Model  OFDM symbols are organized into a number of physical RBs consisting of 12 contiguous sub-carriers 7 consecutive OFDM symbols bandwidth of 10 MHz 50 RBs are available for data transmission 4

5. System Model 5

6. System Model: Deployment 6

7. System Model: Path loss 7

8. System Model: Fading channels  The information theoretical limit is established by defining an outage probability  P out that the instantaneous mutual information for a given fading instance is smaller than the spectral efficiency R associated with the transmitted packet 8

9. Proposed Resource Allocation Algorithm  In this paper vision, femtocells should be invisible in terms of interference generated to neighbour cellular users.  Femtocells deployment presents a very challenging issue: while H-BSs power consumption and interference range should be small, the coverage range at which UEs can meet their QoS constraints should be large. 9

10.  Step 1: [Classification of Interferers] H-UEs overhear the broadcast channel (BCH) and estimate which neighbour HBSs are strong interferers.  Step 2: [Feedback to H-BS] H-UE feedbacks to the H-BS its QoS constraints the momentary Channel State Indicator (CSI) and the cell-IDs of the H-BSs perceived as strong interferers  Step 3: [Feedback to Control Unit] Each H-BS reports this information to the Control Unit (CU). CU stores the set of its neighbours V i. Proposed Resource Allocation Algorithm 10

11.  Step 4: [Computing Scheduling Matrices] CU computes scheduling metric λ i j for every user i on every RB j. RRM ghost implements a Proportional Fair based scheduler, that is RRM ghost uses this metric to build the scheduling matrices M Tx and M rep M Tx : scheduler allocates to each user the minimum number of RBs that meets QoS and power constraints. M rep : to allocate to the served users additional available RBs. Proposed Resource Allocation Algorithm 11 RB 1 RB 2 RB 6 RB 16 RB 9 U1U1 U2U2 U3U3 U1U1 U2U2 U3U3 RB 1 RB 2 RB 16

12.  Step 5: [Scheduling] The CU selects for each user to serve, the minimum number of RBs that meets QoS and power constraints.  Step 5-a: selects the best user-available RB pair (i,j) with the best metric in M Tx  Step 5-b: Each user-available RB pair (i,j), the algorithm checks the set of RBs allotted to user i (RB i ) and selects the highest possible Modulation and Coding Scheme (MCS i ), accordingly. Proposed Resource Allocation Algorithm 12 RB 1 RB 2 RB 6 RB 16 RB 9 U1U1 U2U2 U3U3

13.  Step 5-c: The controller estimates the sum of the Mutual Information I given by set RB i and MCS i. Proposed Resource Allocation Algorithm 13 RB 1 RB 2 RB 6 RB 16 RB 9 U1U1 U2U2 U3U3

14.  Step 6: [MCS Scaling] Given the set of RBs (RB i ) allocated each served user i, the algorithm finds the MCS* of the minimum order that meets the QoS target. If MCS* is different from MCSi, the MCS of user i (MCS i ) is set equal to MCS*.  Step 7: [Repetition] The CU allocates unused RBs to repeat the original message and improve the transmission robustness.  Step 7-a: The scheduler selects the user-available RB pair (i,j) that has the best metric in M Rep.  Step 7-b: If I < R tg, repetitions would cause outage, hence the values of the row corresponding to user I in MRep are set to zero. When I ¸ Rtg, the original message is repeated in the additional RB and MRep(i; j) as well as M Rep (k; j), where k ∈ V i, are set to zero. Proposed Resource Allocation Algorithm 14

15.  Step 8: [Power Scaling] The algorithm estimates the SINR perceived at each served user and reduces the allocated transmission power to meet the SINR threshold given by the target PER and the selected MCS.  Step 9: [Message Reception] Finally, each user collects the information received in each of its allotted RBs and combines these RBs using the chase combining scheme Proposed Resource Allocation Algorithm 15

16. Simulation 16

17. Conclusions  RRM ghost, a novel radio resource management scheme that efficiently uses the available wireless spectrum in a two-tier network.  It limits the undesired effects of interference by reducing the power budget P T required at femtocells to meet target QoS constraints. 17

18. 每周一句  On the contrary, in Scenarios Traf:3 and Traf:4 under RRM ghost, H-UEs achieve performance beyond 90% of Ttg.  on the contrary: 恰恰相反  in contrast: 一定是相反，非常不一樣即可  on the other hand: 換句話說 18