The Impact Of Finite Rate Feedback On The Performance Of Chunk-based Multiuser OFDM Systems

OFDM (Orthogonal Frequency Division Multirths a high speed data transmission technology. It not only transmits th(?)gh-speed serial data stream into relatively low-speed multiple parallel dat(?)eam, but also takes different modulation modes on different subcarriers. This new technology greatly expands the pulse width of the symbols and improves the signal resistance to multipath fading and other harsh transmit conditions.Currently multiuser OFDM resource allocation problems have been studied extensively. It includes the allocation method to minimize transmission power under rate constraints, the allocation method to maximize system throughput under power constraints and the allocation method to maximize weighted average system throughput under power constraints based on fairness. The schemes mentioned above treat single subcarrier as the basic unit in resource allocation. In a broadband wireless frequency-selective channel, different subcarriers experience different channel fading. However, channel gains over contiguous subcarriers are highly correlated for each user. So in practical OFDM systems, a proper number of adjacent subcarriers are grouped into a chunk and a chunk of subcarriers is regarded as the minimum unit for subcarrier allocation. That is, one chunk can only be allocated to one user. To reduce the complexity of the resource allocation and the feedback load, the chunk-based resource allocation has been applied to OFDMA systems. Existing resource allocation scheme have not combined the limited feedback scheme with the resource allocation based on grouping the adjacent subcarriers into chunks. If we feed the channel information for all subcarriers back to the base station, the amount of the feedback may be very large and it’s a waste of network transmission resources.In this paper, we consider the Weighted Sum Rate Maximization (WSRmax) problem with power constraint in multiuser OFDM systems. The adjacent subcarriers are grouped into chunks and the number of subcarriers in every chunk is equal. A chunk is considered as one unit and all subcarriers within one chunk are allocated to the same user. The power allocated to each subcarrier within one chunk is the same. With the above system constraints, the WSRmax problem can be formulated as a non-convex optimization problem. However, we can transform the optimization problem to a convex one by relaxing some requirements. Then the Lagrange decomposition method can be used to solve it. The weighted sum rates of multiuser OFDM systems with three different feedback schemes are compared. The CSI(Channel State Information) that BS can get are different among three feedback schemes. In the first scheme, the accurate CSI of all the subcarriers is known by the BS. In the second scheme, BS can just obtain the quantized values of CSI of all the subcarriers. In the third scheme, only part of the quantized CSI of all the subcarriers is known by the BS and the other CSI is calculated by linear interpolation.