Novel Optimized Cross-Layer Design with Maximum Weighted Capacity Based Resource Allocation for AMC/HARQ Wireless Networks ()
1. Introduction
In order to enhance the throughput of the wireless communication system, link adaptation (LA) technologies are considered not only at the physical layer, but also at the upper-protocol-layers such as data link layer when designing the wireless networks. LA technologies use the instantaneous channel state information (CSI) to adaptively control the data transmission of wireless channel, maintain constant transmission power to reduce the interference of other users and satisfy different business’ needs, and save resources to improve overall throughput for the system. In addition, the adaptive system can easily provide services with different qualities, such as higher information transmission rate, lower packet error rate, and higher spectral efficiency. LA technologies mainly include adaptive modulation and coding (AMC) at the physical layer (PHY) and hybrid automatic repeat request (HARQ) at the data link layer (DLL). The AMC schemes can be used to match transmission rates to timevarying channel, in order to obtain maximum throughput and improve spectral efficiency [1]. Cross-layer design (CLD) which combines AMC and HARQ is extensively studied in order to match the transmission rates to timevarying wireless communication channel conditions [2].
HARQ provides a good tackling technique to solve the tradeoff between channel coding and packet retransmission. It is a well suited solution to enforce the link quality at the medium access control (MAC) layer in wireless environments, and has been adopted in some towards-4G standards (3GPP LTE Release 8, for instance) [3]. It also can be considered as a combination of forward error correction (FEC) and automatic repeat request (ARQ) which is used to improve reliability [1].
A major concern in data communication is how to control transmission errors caused by channel noise so that error-free data can be delivered to the user. A solution to this problem is using HARQ schemes:
Type-I HARQ, the ARQ adopted in 3GPP (R99) is called as HARQ type I. It uses a fixed rate FEC code along with ARQ. In this scheme, both error detection and FEC bits are added to each packet prior to transmission, when the received code word is detected in error, two situations may arise. If the number of errors is within error-correcting capabilities of the code, the errors are corrected. Otherwise, the received coded data block is discarded and a retransmission is requested by the receiver, similar to standard ARQ [4]. The disadvantage of the type I HARQ scheme is that once the code rate is fixed, all parity bits for error correction are transmitted even if they are not all needed, thus reducing the channel use efficiency [5].
Type-II HARQ, it’s an incremental redundancy (IR) HARQ. It considers the time selective property of the wireless channels. At first transmission, the data packet has no or little redundancy. If it fails, each retransmission includes additional redundancy bits from channel encoder; the redundancy is increased in the repeated transmission of the same packet. Thus, retransmission of type-II HARQ only uses the information bits that have not been transmitted in the previous transmissions. If all the channel coded information bits have been transmitted, then the channel code will be retransmitted. When the decoder has received the information packet, it combines all the information packets together to decode. The retransmission packet could not decode alone, it could be decoded only if it was combined together with the previous transmitted information packet [2].
Type-III HARQ, it also an IR scheme but it uses the concept of self-decodable codes; it can be combined with the error data stored and then decoded. The retransmission uses the puncture matrix which is complementary with the puncture matrix for the first transmission. Every puncture matrix has the same code rate and the same error-correction ability [2]. At the receiver, if a frame is detected to be in error, it is stored and a negative acknowledgement (NACK) is fed back. Upon arrival of the retransmission, the receiver attempts to decode the second transmission of the frame. If the decoding is successful, an ACK is fed back; otherwise the frame is stored and combines all the packets together to decode. The retransmission will continue until the decoding is successful or maximum number of transmission has reached [6].
HARQ in 3GPP Long Term Evolution (LTE), both LTE and LTE-Advanced (LTE-A) use HARQ. In LTE, HARQ reordering is done in the radio link control (RLC) layer. So, it operates into different modes: the unacknowledged mode (without ARQ) or in acknowledged mode (with active ARQ). In LTE-A, HARQ is very similar to HARQ in LTE, only small adaptations have been done to cope with the change in the supported bandwidth since LTE-A is supposed to support bandwidth up to 100 MHz [7].
In wireless networking, quality of service (QoS) plays a crucial role in performance measurement. There are four main schemes that are carried out for dynamic subcarrier and power allocation to different OFDM systems users, which allow a flexible multiuser access and enhancement of the multiuser diversity [2]: maximum capacity based resource allocation, proportional fairness based resource allocation, adaptive subcarrier and power allocation in OFDM based on maximizing utility and maximum weighted capacity (MWC) based resource allocation.
The AMC/HARQ CLD concept has been proposed in many literatures such as [1,2,4,6]. In [1], the performance of type I, II, III HARQ with AMC over Nakagami-m fading channel was analyzed, a closed form expression of PER and average spectral efficiency were derived and the conditional probability density function of signal-tonoise ratio (SNR) was calculated using Marcov model. In [2], the HARQ (type-II, type-III) was applied to the cross-layer framework in order to satisfy the prescribed delay and PER constraints at LL. In [4], Jaume Ramis had proposed an analytical link level queuing model to formulate a cross-layer design conceived as a constrained optimization problem to exploit the joint impact on QoS performance of both AMC at the PHY layer and HARQchase combining based-error control at the DLC-layer. Feijn Shi studied the joint design of type-III HARQ protocol at the data link layer and AMC scheme at the PHY layer through closed-form expressions for throughput, average delay and packet loss rate in [6]. Cross-layer scheduling techniques for multiuser OFDM system were introduced in [8,9] by Nan Zhou: in [8], a packet batch based cross-layer scheduling scheme was proposed. This scheme considers the differences between the batches in the same queue, so it is more flexible and efficient than conventional queue based scheduling. In [9], cross-layer resource allocations based on MWC technique and delay satisfaction (DS) scheduling scheme in the downlink multiuser OFDM system were proposed. The current paper will try to investigate the CLD and its enhancements for OFDM based wireless networks. Differently from the previously mentioned publications, the current paper introduces another vision for CLD in both uplink and downlink multiuser OFDM system by considering the packet error rate (PER) as well as the SNR constraints on the overall performance.
In this paper, based on the AMC/HARQ cross-layer design, the system capacity was investigated for different SNR of OFDM channels for different HARQ schemes. Also, a novel PER based data scheduling technique based on MWC resource allocation was proposed.
In Section 2, the system model is presented. Sub-carrier and power allocation are presented in Section 3. PER based data scheduling is presented in Section 4. Numerical results and analysis are shown in Section 5. Finally, Section 6 concludes this paper.
2. System Model
The point-to-point wireless packet communication system under consideration in this paper is illustrated in
Figure 1, showing the most important blocks involved in the transmission and reception processes for a single user. The HARQ protocol with AMC is used over the Nakagami-m fading channel, the channel gain is assumed to be constant during the transmission of one frame, and it varies during the transmission of the next frame. Also, the channel is assumed to support QoS-guaranteed traffic characterized by a packet error rate (PER) and a predetermined acceptable link layer PER Ploss. The processing unit at the data link layer is a packet of fixed size, and the processing unit at the physical layer is a frame made of a variable number of packets that depends on the transmission mode (TM) selected by the AMC scheme. The receiving buffer can accurately detect the received SNR and decide the next modulation scheme, then feedback channel state message to the source node. On the other hand, if the destination node cannot decode the received data packet correctly, it feeds back NACK signaling to the source node, otherwise, it feeds back acknowledge (ACK) signaling, and the transmission of all signaling messages are assumed to be error free, the maximum number of retransmissions is denoted Rmax.