中繼式正交分頻多工存取系統之公平動態資源分配

Abstract

正交分頻多工存取(OFDMA)網路的系統容量和涵蓋範圍可透過動態、機會式的資源分配來大幅度的改善。然而這一類的無線資源分配牽涉到眾多系統參數與設計之選擇及實務考量，複雜度相當高，絕大部分情況下無法有最佳的解決方案。本文遂只考量於單一基地台、多個合作式中繼台和移動台的細胞式通訊系統下的實際可行的次佳解。我們考慮的信號格式是類似IEEE 802.16e所使用的分時多工，且只討論上傳的無線資源分配。這些無線傳送資源包含載波、電力和中繼台。中繼台可能是專用的或者是沒有信號傳送，暫時閒置的移動台。除了第五章討論的是放大再轉送（amplify and forward）之外我們都假設中繼方式是所謂的解碼再轉送（decode and forward）。

我們首先考慮的系統設計課題是：盡量降低總傳送能量但需滿足傳送率、服務品質及每一次載波上所能攜帶的位元上限之要求。服務品質是指傳送錯誤率的大小，這與次載波上所承載的位元量和所需要的傳送電力大小是有一定關係的。因此位元分配(速率分配)就決定了能量的需求。第二項探討的課題則是在兼顧公平性的要求下使得傳送速率總和最大，並要滿足每個用戶之電力、服務品質和最低傳送率要求。

對於以上兩項課題我們個別都提出了兩種線性複雜度的次佳分配法。電腦模擬結果顯示：我們所提出的演算法皆有甚佳的效能表現，除了達到大量能量的節省或者是接近最佳的傳送率總和之外，我們也能兼顧維持穩定且良好的用戶間之公平性。

Capacity and coverage of an Orthogonal Frequency Division Multiple Access (OFDMA) network can be greatly enhanced by dynamically and opportunistically allocate the radio transmission resources. We restrict our investigation to a single-cell system with multiple cooperative relay stations and mobile stations (MSs). A TDD scenario is assumed and only the uplink transmission with the base station (BS) handling the resource allocation is considered. The transmission resources include subcarriers, power and relays with the later being dedicated relay stations or cooperative MS's with unused signal dimensions.

We first consider the scenario that the total transmit energy is to be minimized under rates, QoS and maximum per-subcarrier loaded bits constraints where QoS refers to the bit error rate (BER) requirement. As there is a deterministic relation between the number of bits carried by a subcarrier and the power (energy) needed to achieve a desired BER performance, once the QoS requirement is given, bit-loading (rate-assignment) is equivalent to energy appropriation. The second scenario is concerned about the problem of sum rate maximization with a fairness consideration plus power, QoS and minimum rates constraints.

For both scenarios we present two linear-complexity suboptimal solutions. Numerical results are given to show that the proposed solutions do offer attractive performance advantages of either energy-saving or near-optimal sum-rate while maintaining much improved and robust fairness performance.