System Model of OFDMA
Chapter 2 System Model of OFDMA
2.1 OFDM Transceiver
Fig. 2.1 OFDM Transceiver for Single User
igure 2.1 shows a simplified OFDM transceiver block diagram. An OFDM transmitter separates a symbol of serial bits within symbol time Ts into a parallel form; i.e., each bit is transmitted within the same Ts and fed into the corresponding subcarrier.
In the discrete time domain, an OFDM modulator can be easily implemented by inverse fast Fourier transform (IFFT) and a parallel-to-serial (P/S) converter. In order to cancel the inter-carrier interference (ICI) of a subchannel due to multipath, a cyclic prefix (CP) is attached, which is a copy of the last portion of OFDM signal, to the front (or head) of itself.
In contrast to OFDM transmitter, an OFDM receiver removes the CP first, and then converts serial bits into parallel form for performing the fast Fourier transform (FFT) as a demodulator.
Finally, the output bits from the FFT block will be fed into the de-mapper block to restore the
F
original bits, and P/S conversion is followed.
2.2 OFDMA System
In the multiuser environment, the OFDMA system is a time-frequency multiplexing system. Users are multiplexed both in time and frequency domain, with pilot and signaling information. In the frequency domain (i.e. the sub-carrier domain), users data symbol can be multiplexed onto different numbers of useful sub-carriers. In addition, sub-carriers or group of sub-carriers can be reserved to transmit pilot, signaling or other kinds of symbols.
Multiplexing can also be performed in the time domain, as long as it occurs at the OFDM symbol rate or at a multiple of the symbol rate. The modulation scheme (modulation level) used for each sub-carrier can also be changed at the corresponding rate, keeping the computational simplicity of the FFT-based implementation. The 2-dimensional time-frequency multiplexing scheme of OFDMA system is depicted in Fig. 2.2 [7].
Frequency (Useful sub-carriers) →
D4 D4 D4 D4 D4 D6 D6 D6 D6 D6 D6 ↓
D4 D4 P D4 D4 D6 D6 P D6 D6 D6 TTI
D4 D4 D4 D4 D4 D6 D6 D6 D6 D6 D6 ↑
D2 D2 D2 D2 D2 D2 D2 D2 D2 D2 D2
P D2 D2 D2 P P D2 D2 D2 P D2
← Time
D2 D2 D2 D2 D2 D2 D2 D2 D2 D2 D2
P = pilot or signalling, D = data. TTI = transmission time interval
The subscript indicates the modulation level M=2,4 or 6 (QPSK, 16QAM or 64QAM).
Fig. 2.2 An Example of OFDM 2-D Structure
According to [7], three types of OFDM physical channels are defined: OFDM common pilot channel (OFDM-CPICH), OFDM shared control physical channels (OFDM-SCCH), and OFDM-physical downlink shared channels (OFDM-PDSCH). OFDM-CPICH contains pilot information, where the pilots are inserted in the time-frequency domain and must satisfy the 2-D sampling theorem in order to enable reconstruction of the time-and frequency varying channel response [8]. OFDM-SCCH, contain signaling information. The specific frequency locations used for signaling should be scattered, in order to benefit from frequency diversity. Conversely, the time locations can be spread across the subframe, while limited to a small number of OFDM symbols to ease extraction of the OFDM-SCCH information by users. OFDM-PDSCHs, are dedicated to carry data or higher layer signaling information.
The concept of downlink OFDMA system is illustrated in Fig. 2.3, where K users and N subcarriers are assumed. At the transmitter of a base station (BS), the Bits and Subcarrier Allocation function block executes every transmission time interval (TTI) to distribute all users’ data streams to each subcarrier with appropriate modulation order. Its functionality is performed according to a specific RRA algorithm so that the QoS requirements of all active users are fulfilled, or the overall throughput of the OFDMA system is improved. If a user’s data stream cannot be transmitted completely at a TTI, the data stream then is queued in a buffer and waits for the RRA in the following TTI. Notably, the queued data are treated differently for RT and NRT users due to their delay tolerances. The data stream of a RT user will be dropped if it excesses the maximum delay tolerance of the QoS requirements, while the data stream of a NRT user will not. The Adaptive Modulator at each subcarrier maps the data bits from the previous function block, Bits and Subcarrier Allocation, into a QAM symbol. Finally, the last function block, OFDM Modulator, will converts the N QAM symbols into an OFDM symbol by IFFT.
At the kth user’s receiver, OFDM Demodulator converts a noisy OFDM symbol to N QAM symbols by FFT. It is assumed that the resource allocation information of the kth user is obtained in advance, which includes the information subcarriers allocations and the
related modulation orders. With these information, the OFDM Demodulator filters the needed QAM symbols, and then Adaptive Demodulators converts them to extract the original data bits. Finally, the last function, Bits and Subcarrier Extraction, will restore all the data bits to the kth user’s data stream.
Fig. 2.3 Downlink OFDMA System
2.3 System Constraints
In the downlink OFDMA systems, some realistic considerations are taken into consideration in this thesis. Here, four system constraints are imposed on the design of radio resource allocation algorithm for Bits and Subcarrier Allocation at the transmitter of a BS, which are stated as follows.
(I) Subcarrier Allocation Constraint
In OFDMA system, a subcarrier can only be allocated to one user at a TTI. Denote Ψnthe index of the user granting the nth subcarrier and Φnthe modulation order assigned to the nth subcarrier.
Ψn =
{
φ,1,L,K}
, for n=1,L,N (1) Φ =n{
0,1, 2, 3 (no transmission, QPSK, 16QAM, 64QAM) , for}
n=1,L,N (2)(II) Maximum Subcarrier Power Constraint
To prevent from the peak-to-average power ratio (PAPR) problem, the power allocation of a subcarrier should be limited, which denotes as the maximum subcarrier power constraint,
pmax.
(III) Total System Power Constraint
The power allocation for data transmission should be limited at the transmitter of a BS;
that is, the sum of the maximum power allocated to each subcarrier will be bounded at the
constraint is formulated as follows, allocation buffer constraint is formulated by
{
k}
RkB k Kwhere RkB denotes the required rate to clean up the packets of the kth user’s buffer.
2.4 Channel Model
In wireless environment, the bandwidth of each subcarrier should be smaller than the coherence bandwidth of the channel to overcome the frequency-selective fading. No correlation between each subcarrier is assumed, and each subcarrier has a flat-fading channel with a constant link gain within a TTI, which is given by
αk,n = FL,k ⋅FS,k,n, (7) , (8)
whereα denotes the link gain of the kth user at the nth subcarrier. The link gain consists of k ,n two parts: one is large-scale fading, and the other is small-scale fading. FL,k is large-scale subchannel with Rayleigh distribution due to multipath. In a mobile, the large-scale fading of all subchannels are the same at one random time instant, but the small-scale fading of all
subchannels are different.
Furthermore, perfect estimation of the CSI of each users is assumed, and retransmission due to transmission error is not considered in this thesis.
2.5 Services
In this thesis, two kinds of service are considered, one is real-time (RT) service and the other is non-real-time (NRT) service. The RT services are delay-sensitive, such as voice services, and their traffic characteristics are described by ON/OFF model. The packets are generated in a constant rate during ON period, while no packets are generated during the OFF period.
Four QoS requirements are considered for the RT services. The first requirement is required transmission rate, R*during the ON period, the second requirement is required bit error rate, BER , the third requirement is maximum delay tolerance, * D , and the last * requirement is maximum packet dropping ratio, P . The packets of RT services will be D* queued in buffers if the radio resources are not enough for transmission; however, they will be dropped if the delay of head of line (HOL) packets exceeds the maximum delay tolerance.
A RT service is called “unguaranteed” if the ratio of dropped packets over the total arrival packets is larger than the maximum packet dropping ratio P . D*
The NRT services are delay insensitive, such as WWW application, and their traffic characteristics are described by Pareto model. The packets are generated with burstness when a NRT user click a web page, and no packets are generated when the NRT user is reading the web page. Different from the RT services, the packets of NRT services can be queued for a long time without being dropped. Only two QoS requirement are considered for the NRT services: required BER, BER* and minimum required transmission rate, Rmin* .
Moreover, token bucket method is implemented in the buffer management of the NRT services. The system generates tokens at a constant rate for a NRT user, and the actual transmission rate of the packets of the NRT user are determined by a RRA algorithm in Bits
and Subcarrier Allocation according to the remaining tokens, which is the difference between the total aggregated tokens and the total transmitted packets.