A Signal Detection Method for Uplink Multiuser MIMO Systems Tae Ho Im, Sungwook Yu, and Yong Soo Cho School of Electrical and Electronic Engineering, Chung-Ang University, Seoul, Korea Abstract — In this paper, we propose an efficient signal detection method for uplink multiuser systems based on collaborative spatial multiplexing (CSM). The proposed method achieves near-optimal performance and shows only 0.8dB loss at the target frame error rate (FER) of 10 -2 . It also requires substantially lower computational complexity than the QRD-M method especially when the modulation order is high (e.g., 16- QAM and 64-QAM). Moreover, the error performance of each user is almost the same in the proposed method, which is an important property in a multiuser multiple-input multiple-output (MIMO) system where each user’s error performance must satisfy some fixed error rate criteria. This is a big advantage over other detection methods such as the ordered successive interference cancellation (OSIC) and the QRD-M methods, where the error performance varies greatly among users. Index Terms — MIMO, Uplink, Multiuser. I. INTRODUCTION MIMO communication systems have received tremendous attention because of their high spectral efficiency and near- capacity performance. As a result, MIMO has become a key component in several wireless communication standards, including LTE-Advanced and IEEE 802.16m [1]. A small- hand held device, however, may not be equipped with several antennas due to its small size and the additional cost incurred. In order to solve this problem, multiuser MIMO schemes have drawn increased attention, where several users (each with a single antenna mobile device) share the same MIMO channel. For example, four mobiles can transmit simultaneously, each on an independent MIMO path. Even though multiple signals are sent on the same frequency, a MIMO base station (BS) can distinguish between different signals if the main energy of each signal arrives from a different direction. Thus, the peak transmission rates and the system throughput rates can be increased by allowing multiple simultaneous in-sector transmissions on the same time-frequency resources [2]. This technique, which is called CSM, is considered as the uplink transmission scheme of the LTE-Advanced and IEEE 802.16m [3]. The success of this technique requires the perfect cancellation of multiuser interference, which poses considerable research challenges.Both uplink multiuser systems based on CSM and spatially-multiplexed (SM) MIMO systems can be modeled in the same way. Thus, any detection methods for SM-MIMO systems (such as the OSIC and the QRD-M methods) can be used for uplink multiuser systems [4],[5]. In this paper, we propose a signal detection method which is especially suitable for uplink multiuser systems based on CSM. The proposed method achieves near-optimal error performance while requiring significantly less complexity than the QRD-M method. The proposed method has another advantage in that the error performance of each user becomes almost the same, unlike other conventional detection methods such as OSIC and QRD-M. Also, the log-likelihood ratio (LLR) values for all the bits can be obtained without using any LLR clipping methods in the proposed method even though it uses only a small number of candidate vectors. The rest of this paper is organized as follows: Section II describes the system model for uplink multiuser systems based on CSM and the proposed signal detection method. Sections III and IV compare several detection methods in terms of computational complexity and error performance. Finally, the conclusions are given in Section V. II. DETECTION METHOD FOR UPLINK MULTIUSER SYSTEMS BASED ON CSM Here, we'll consider an uplink multiuser system with N T receiving antennas and N U transmitting users (each with a single antenna), where N R ≥ N U . The transmitted signal and the received signal are then related as follows: = + y Hx n (1) where 1 2 U N x x x ⎡ ⎤ = ⎣ ⎦ x " is the transmitted symbol vector of the whole group of users, 1 2 R N y y y ⎡ ⎤ = ⎣ ⎦ y " is the received symbol vector in the BS, and 1 2 R N n n n ⎡ ⎤ = ⎣ ⎦ n " is an independent and identically distributed (i.i.d.) complex zero-mean Gaussian noise with a variance of 2 σ per dimension. The matrix H is an R U N N × channel matrix where the element , ij h represents the complex transfer function from the j-th user to the i-th receiving antenna, and all , ij h 's are all i.i.d. complex zero-mean Gaussian with a variance of 0.5 per dimension. We will assume that each user employs an independent modulation and coding scheme, and each component i x in x is independently drawn from a complex constellation such as QAM, 1, 2, , i U i N Ω - = " denotes the set of all constellation points for the i-th user, and i Ω denotes the number of elements in the set. As can be seen from (1), the system model for an uplink multiuser system based on CSM is the same as that for a SM-MIMO system. In an uplink multiuser system, it is important to guarantee the worst-case transmission quality for each user. We will assume that, in the base station, the transmission quality of each user 978-1-4577-0963-0/11/$26.00 ©2011 IEEE