An Efficient Scheduling Scheme for Resource Utilization and Performance Enhancement of Multi User MIMO Wireless Systems

INTRODUCTION

Each antenna element on a MIMO system operates on the same frequency and therefore does not require extra bandwidth. Also, for fair comparison, the total power through all antenna elements is less than or equal to that of a single antenna system. Hence MIMO system consumes no extra power due to its multiple antenna elements (Tsoulos, 2006). As a consequence of their advantages, MIMO wireless systems have captured the attention of international standard organizations. The use of MIMO has been proposed multiple times for use in the high-speed packet data mode of third generation cellular systems (Molisch, 2005; Gesbert et al., 2003). Also, to increase the multiplexing gain, the system capacity has to be increased. This can be achieved by using multiple antennas for data signals transmission. Signal processing at the transmitter and at the receiver may result in the effect of coherent combining hence the Signal to Noise Ratio (SNR) of the system is increased (Foschini and Gans, 1998; Tse and Viswanath, 2005; Chanthirasekaran and Bhagyaveni, 2009). Spatial diversity using Space-time Block Coding (STBC) has been actively investigated by Zhang and Gulliver (2005), Dohler and Aghvami (2004) and Perez et al. (2005). Zhang and Gulliver (2005) derived a closed-form Symbol Error Rate (SER) of the STBC for various modulations and fading channels; the outage probability of the STBC over Nakagami fading channels was derived by Dohler and Aghvami (2004). Perez et al. (2005) presented a closed-form approximation of the STBC for capacity in various fading channels. For a reliable multimedia communication service, Spatial Multiplexing based MIMO technique plays a vital role. The spatial diversity is exploited if the same signal is transmitted on all antennas and can be used to increase the reliability of reception while the spatial multiplexing gain is achieved by transmitting different signals on each antenna to increase the throughput for a fixed reliability level (Tse et al., 2004). It has been shown that it is a trade-off between the spatial diversity gain and the spatial multiplexing gain of a MIMO-system (Tse et al., 2004). To jointly exploit the gain from Multi User Diversity (MUD) and multiple antennas in MIMO systems, several users have to be scheduled in each time-slot (Jagannathan et al., 2006). The spatial diversity in a MIMO system can be exploited by using e.g., Space-Time Block Coding (STBC) to obtain better error production (Alamouti, 1998). In MUD environment, combination of scheduler and antenna diversity schemes is analyzed (Jiang et al., 2004; Chen and Wang, 2006). Chanthirasekaran and Bhagyaveni (2011) presented a scheduler with STBC for BER improvement. Combining the scheduler with STBC for the improvement of the multi user packet delivery, challenges of fairness and resource utilization was not addressed in those studies.

In this study, an efficient scheduling is proposed. This scheduler allocates resource adaptively based on Signal to Noise Ratio (SNR) of the multiple users for performance improvement and verify the resource utilization. The scheduler grants resource for the set of best users under poor performance region the threshold based fair scheduling (TFS) and grant resources for the multiple users whose SNR is above threshold under good performance region of TFS. The performance of the proposed scheduler is compared with other scheduler policies like Fairness Scheduling (FS), Priority Scheduling (PS) and threshold based fair scheduling (TFS). From the results it is observed that ES scheme produces better PDR.

SYSTEM OVERVIEW

Here, the model of downlink multi user MIMO system where a single Base Station (BS) with ‘r’ pair of transmit antennas communicates with ‘n’ number of mobile users each with M_R receive antennas are shown in Fig. 1.

Figure 2 shows functional blocks of the base station. Here, the n users place their request to the BS using request to send (RTS) packets. The scheduler using proposed scheduling algorithm grants resources to the multi users using Clear to Send (CTS) packets. Then these granted users transmit their message using STBC coder under MIMO system. In a SISO system they employ single antenna for transmission.

The proposed scheduling algorithm allocates resources effectively for multiuser. The various multiuser scheduling algorithms are compared with proposed scheduling algorithms.

SCHEDULING ALGORITHMS

We assume that the base station knows the channel state information. It can select a group of users from all the requested users to achieve the better performance with the help of various scheduling. Here the equal number of transmit antennas (one for SISO and two for MIMO) are allotted per user in the time instant t_k for n_tk number of requested users by considering the number of r_tk resources available.

Fig. 1:	Model of downlink multi user MIMO system

Fig. 2:	Block diagram illustrating the internal blocks of base station of MU-MIMO system

The following scheduling algorithms are used in the scheduler at the base station for performance comparison.

Fair scheduling: Let n_tk be the total number of users who placed request in a time slot t_tk._. This scheduler grants resources r_tk based on First Come First Serve (FCFS) basis. If the request received at BS in a time slot t_k is {x_t1, x_t2, x_t3,..., x_tk} where x_t1 is the user whose request is received at time t₁, x_t2 is the user whose request is received at time t₂ analogously x_t3,... x_tk where t₁<t₂<t₃<.....<t_k with Σ x_tk = n_tk. The granted users at time slot t_k as given in Eq. 1 receive CTS from the base station:

(1)

where, t₁<t₂<t₃<.....<t_P<....<t_k.

Algorithm for fair scheduling is as follows

As fair scheduler grant resource for the number of user based on first come first serve basis, it maintains the service fairness but the BER performance of the system is about 10^-2 as shown in Fig. 3. In order to improve the BER performance PS is modeled.

Priority scheduling: The base station using its Channel State Information (CSI) computes SNR of each requested user. Priority scheduler sort the n_tk users based on their computed SNR strength. The scheduler grants resource to first r_tk sorted users. The granted users G_tk at time slot t_k is given as:

(2)

Where:

Algorithm for priority scheduling is as follows:

As priority scheduler grants resources for ‘r’ number of best user, the performance of the system is good if r<<n. The performance decays when the number of resources is closely equal to number of users.

Threshold based Fai scheduling: Threshold based fair scheduler computes offer-able SNR threshold ‘Th’ by using average of minimum SNR and maximum SNR of requested user as given in Eq. 3:

(3)

Fig. 3:	BER performance comparison ES with TFS, PS and FS (1X1 System)

Then r_tk resources are granted to the users in the time slot t_k based on first come first serve and if their SNR_i≥ ‘Th’ for I = 1, 2….n_tk. Let x = {x_t1, x_t2, x_t3,..., x_tk} be the users request received at time slot t_k and Let x' = {x'_t1, x'_t2, x'_t3,..., x'_tk} be the users whose SNR’s are greater than or equal to ‘Th’ where x'⊂x. The granted users in the time slot t_k is given as:

(4)

Algorithm for Threshold based fair scheduling is as follows:

TFS scheduling gives better performance when the number of resource exceeds about fifty percent of number of users. When the number of resource is less than or equal to fifty percent of number of users, the new Efficient Scheduling (ES) algorithm is developed for enhancing the system performance.

Efficient scheduling: Efficient scheduler works under two phase, the network learning phase and resource granting phase. In network learning phase the scheduler runs various scheduling algorithms and evaluates the performance of the network under various load condition. The results are maintained in performance table. During resource granting phase, as and when the request arise, the scheduler enables the Scheduling Algorithm Selection Process (SASP) which checks the performance table to identify the best algorithm for the present request condition and adapts the suitable scheduling algorithm. The granted users G_tk at time slot t_k is given as:

(5)

Algorithm for this scheduling is as follows:

SIMULATION AND RESULTS

The system is modeled using one base station receiving request from n_tk users. The users demand resources from the base station by using request packet. The performance of scheduling is simulated for two cases. In case 1 each user has one antenna and demands one antenna resource. In case 2 each user has 2 antennas and demand 2 antenna resources from the base station. In 2x2 MU-MIMO systems the user data’s are transmitted after Alamouti-Space Time Block Coding (STBC). Simulation parameters are shown in Table 1.

Figure 3 shows the BER performance of 1X1 antenna system with different scheduling. It has been observed that the ES outperforms well as compare to TFS and PS. From Fig. 3, it has been observed that this system is able to achieve the bit error rate in the order of 10^-3. To improve the BER performance 2X2 MU-MIMO system is derived and their performance is shown in Fig. 4 and 5.

The performance table during learning phase of efficient scheduler is illustrated in Table 2.

Figure 4 gives BER performance and it shows that the threshold based fair scheduling performes well as compared to priority scheduling when the number of resources exceeds fifty six percent of number of users otherwise PS gives better performance because it serves group of best users. When the available resources increases and become closer to the number of users, priority scheduling gives same performance as FS.

Also it is observed that the efficient scheduling is able to improve the BER performance as compare to threshold based fair scheduling when the number of resource is less than 56% of number of users.

Fig. 4:	BER performance comparison of 2X2 ES with other scheduling

Fig. 5:	PDR performance comparison of 2X2 ES with PS and TFS

Table 1:	Simulation parameters for MIMO-STBC system

Table 2:	Performance table of ES

Also it gives better performance as compare PS when the number of resources exceeds fifty six percent of the number of users.

The Packet Delivery Ratio (PDR) of this scheme is shown in Fig. 5. The PDR performance of ES scheme is found improved by ten percent as compared to the TFS scheme and same as compare to PS till the number of resources are fifty six percent of the number of users. As it exceeds ES outperforms well and the PDR performance of PS rasticly decreases under this case. It is also observed that the ES is able to maintain the good PDR performance.