Subcarrier Allocation Algorithms for multicellular OFDMA networks without Channel State Information I.N. Stiakogiannakis, D.A. Zarbouti, G.V. Tsoulos, D.I. Kaklamani, NTUA Department of Telecommunication Science and Technology, UoP Contact e-mail : istiak@esd.ntua.gr 1
Scope of the paper Simulation and comparative analysis for four subcarrier allocation algorithms Coordinated Sequential Random Random-Coordinated, a cell-splitting technique that uses both coordinated and random subcarrier allocation Common characteristic of these algorithms is that they do not require CSI knowledge at transmitter end 2
OFDMA RRM Three-step approach Base Station Selection The BS that shows the best channel gain for the central network frequency is selected to serve the new user Subcarrier Allocation Specification of the number of subcarriers to be allocated If user asks for R b u bps while system uses 2 -QAM modulation with subcarrier spacing Δf Hz, the number of subcarriers is s u =ceil(r u /(bδf)) Allocation of specific subcarriers to user Subcarrier Allocation Algorithms Bit Loading and Power Control Definition of the transmit power of the user and modifications on transmit power of co-channel users in order to preserve QoS (Bambos-like power control) 3
Subcarrier Allocation Algorithms Sequential Subcarrier Allocation BSs allocate subcarriers sequentially, one after another Coordinated Subcarrier Allocation BSs with odd ID allocate subcarriers sequentially with a top-down logic while BSs with even ID allocate subcarriers with a bottom-up logic Random Subcarrier Allocation BSs select subcarriers to allocate randomly from the set of available subcarriers Random-Coordinated Subcarrier Allocation Cell is split into two regions, internal and external, with a gain threshold criterion Internal region employs random subcarrier allocation External region employs coordinated subcarrier allocation 4
Simulation Setup Number of Cells 19 Cell Radius 530m Central Frequency 2.5GHz Available Bandwidth No of Subcarriers 128 10MHz Subcarrier Spacing 78.125kHz Modulation Bit Rate Classes QPSK Bit Error Rate 10-5 64... 1024 kbps Propagation Model COST-Hata Gain Threshold -106.5dB Gain threshold is derived as the channel gain at half of the cell radius Monte Carlo Simulations : 250 Iterations Users are located randomly in the geographical area of the network and select their bit rate class randomly 5
Simulation Results 6
Central Cell Bit Rate Offered Bit Rate reaches 20Mbps Sequential shows the worst performance Random is better than Random-Coordinated at 75% of cases (0.3Mbps at 50% of cases) Sequential and Coordinated suffer from the same drawback of sequential allocation 7
Central Cell Transmit Power Sequential algorithm requires less transmit power than other algorithms Coordinated algorithm requires less transmit power than random algorithm (~4dB) though their performance is comparable 8
Multicellular Bit Rate Sequential algorithm shows by far the worst performance The coordinated algorithm triples this performance (175Mbps, 50%) Random algorithm outperforms coordinated (14.5Mbps, 50%) Random-Coordinated algorithm outperforms random (11.5Mbps, 50%) 9
Multicellular Transmit Power Random and coordinated algorithms have comparable requirements on transmit power Random-Coordinated algorithm needs less transmit power while shows slightly better performance 10
Number of Users per Service Class The number of users per service class decreases as the required bit rate increases Random algorithm performs better for bit rate classes up to 256kbps Coordinated algorithm is better for 512kbps & 1024kbps Random-Coordinated algorithm inherits from random algorithm the good performance for lower bit rate classes (slightly worse) and from coordinated for higher bit rate classes(slightly better) 11
Subcarrier Reuse Sequential algorithm : Many subcarriers remain unused Coordinated algorithm : A high peak at nine; at every cell half of the subcarriers are used Random algorithm : Smoother distribution due to interference averaging Random-Coordinated algorithm : Combination of random & coordinated algorithm; the coordinated part (external region) shows worse performance (leftshifted) while the random part (internal region) shows better performance (right-shifted) 12
Mean Reuse Factor Random algorithm is closer to the goal of Reuse-1 Random-coordinated algorithm, although shows better overall performance, results to higher mean reuse factor due to the fact that the external part cannot use effectively the assigned fraction of bandwidth; this is caused by lower channel gain and higher interference level Subcarrier Allocation Algorithm Mean Reuse Factor Coordinated 2.24 Sequential 4.86 Random 2.11 Random-Coordinated 2.42 13
Conclusions Sequential algorithm is proved to be inappropriate for the system configurations examined herein Random algorithm outperforms both sequential and coordinated algorithms in terms of system capacity Coordinated algorithm shows better performance for higher bit rate classes (512kbps & 1024kbps) contrary to random algorithm that performs better for lower classes (64kbps, 128kbps & 256kbps) The proposed cell-splitting scheme, named random-coordinated algorithm, improves further system capacity while requires less transmit power. 14
Thank you 15