An efficient bandwidth sharing strategy through users' cooperation for multirate networks
Noha O. El-Ganainy,
Electronics and Communications department, Faculty of Engineering, Arab Academy of Science and Technology AAST, Alexandria, Egypt
Corresponding author: Noha O. El-Ganainy, Electronics and Communications department, Faculty of Engineering, Arab Academy of Science and Technology AAST, Alexandria, Egypt. (firstname.lastname@example.org)
 This paper adopts an efficient strategy that allocates the network resources to the users according to the transmission rate requirement. Based on cooperation and the use of variable length spreading codes, the terminals requiring high transmission rate are supported by larger bandwidth and power in order to enhance their quality of service. The proposed resource allocation strategy assigns the terminals with low transmission rate to cooperate with those with high transmission rate by dedicating a part of their bandwidth and power to either relay their data or send a part of it. This will result in a slight limitation on the performance of the cooperating terminals (low rate). Two proposed schemes are presented namely the relaying strategy and re-allocation strategy. These strategies are evaluated against the commonly used rate-matching algorithm which constrains the terminals to fairly share the bandwidth regardless of the transmission rate. The proposed receiver is simple and efficient and the utilized spreading codes provide signal separation and limit multiple access interference MAI. In a CDMA-based framework supporting multirate transmissions, a thorough performance assessment in terms of the error probability and the probability of outage is discussed for the considered strategies under different transmission conditions. The proposed strategies enhance the performance of the high rate user compared to the rate-matching algorithm and slightly hold back the low rate user performance. Moreover, the re-allocation strategy enhances the system capability to support higher transmission rate at no additional cost.
 The concept of cooperative transmission has been deeply investigated during the last few years and proved remarkable efficiency in several applications [Aazhang et al., 2003a, 2003b; Hedayat et al., 2004; Laneman, 2002; Laneman et al., 2004; Azarian et al., 2005; Lops et al., 2006]. Recently, it has been proposed in accordance with resource allocation as a solution to the limited resources in different networks [Gastpar et al., 2005; Erkip and Gunduz, 2007]. Among all the limited resources, the bandwidth and power must be carefully used to avoid severe performance degradation. In networks supporting multirate transmissions, the terminals commonly equally share the available resources regardless of their transmission rate which is judged to be a waste of the network resources. In most of these networks such as the Universal Mobile Telecommunications Systems (UMTS), the terminals with different rates are assigned the same bandwidth size and a fitting variable length spreading code [Ottosson and Svensson, 1995; Chen and Hwang, 2006; Dell'amico et al., 2002]. This is called the rate-matching algorithm. This bandwidth assignment results in very large spreading factors, unnecessarily, for terminals with low rates and very small spreading factor, which holds back the performance, for terminals with high rates.
 In a CDMA-based transmission, this paper seeks to enhance the performance of the multirate transmissions through users' cooperation. The available resources are efficiently allocated to each user according to his/her transmission rate without additional power consumption at the cooperating terminal. The low rate terminal utilizes a part of his bandwidth to transmit his own data and the remaining part is dedicated to cooperate with the high rate terminal either using the relaying strategy or the re-allocation strategy. According to the relaying strategy, the low rate user sends a copy of the high rate user's data on a part of his bandwidth while following the re-allocation strategy he sends a part of the high rate user's data. We introduced an overview on our early work on the considered cooperative strategy in [El-Ganainy and El-Khamy, 2011].
 In a multirate CDMA-based framework and using variable length spreading codes, the proposed cooperative strategies are applied and evaluated against the use of the rate-matching algorithm. A thorough performance assessment is performed in terms of the error probability and the probability of outage to evaluate both strategies using different transmission rates and under different channel conditions. The proposed strategies are shown to provide better quality of service to the high rate user with a slight impact on the low rate user's quality of service. Moreover, they are found to be more robust to channel variation due to cooperation and capability to support higher transmission rate with no additional cost. The rest of this paper is organized as follows: section II highlights the resource management in a multirate network, section III carefully discuss the proposed cooperative strategies, section IV observes the simulation results comparing the proposed cooperative strategies with the rate-matching algorithm under different transmission conditions, and section V provides some concluding remarks.
2 Multirate Transmissions: Conventional Solutions
 Wireless networks supporting terminals with multirate requirements normally assign the same bandwidth size to all terminals which limits the quality of the high rate data compared to the low rate data [Ottosson and Svensson, 1995; Dell'amico et al., 2002]. This is realized by assigning variable length orthogonal spreading codes to fit the different rates to the same bandwidth size, namely called the rate-matching algorithm. Most of the practical systems such as the UMTS commonly use the Orthogonal Variable Spreading Factor (OVSF) codes for the multirate frameworks [Chen and Hwang, 2006; Dell'amico et al., 2002; Darnell and Fan, 1996]. In a CDMA-based transmission and applying the rate-matching algorithm, the high rate user's received signal yH,d at the ultimate receiver and that of the low rate user yL,d are shown in (1).
Where Pi denotes the power transmitted at the source i, hi,d is the Rayleigh channel fading coefficients between the source i and the ultimate receiver d, x is the transmitted symbol with unit power, ni,d denotes the AWGN, H is a suffix that symbolizes the high rate user, and L a suffix describing the low rate user. In order to fit the same bandwidth size, the high rate users are assigned short spreading code lengths while the low rate users are given very long spreading code length. This causes two problems; the small spreading factor assigned to the high rate user has the potential to limit its quality of service. On the other hand, the low rate users are unnecessarily supported by very large spreading factor that provides a quality of service that is much higher than the required level. However, it wastes the bandwidth and the power resources which support the need of an efficient resource allocation in a multirate framework to enhance the overall performance.
3 The Proposed Cooperative Allocation Strategy
 In this paper we propose an efficient allocation of the network resources for multirate transmissions through users' cooperation. The terminals cooperate without the need of additional power consumption compared to noncooperative transmissions. The high rate transmission is supported by a larger bandwidth and power compared to the low rate. Two strategies are employed: relaying and re-allocation strategy. In either case, the low rate user cooperates with the high rates user using a part of its bandwidth and power. The cooperative strategy is proposed for different networks while the upcoming simulations consider a cellular system supporting multirate transmissions. The users are able to cooperate in both the uplink and the downlink to forward each other data. The two proposed strategies apply different duplexing mode: the relaying strategy is based on half-duplex while the re-allocation strategy needs full-duplex for the uplink transmission and half-duplex for the downlink. The cooperative procedure employs a decode-and-forward protocol at the cooperating terminal. In each phase, the base station locates a terminal transmitting at low rate and assigns it to cooperate with a partner with high rate requirement. Two spreading codes with different lengths are required for the high rate user and one code for the low rate user. Once the base station ties up two users as user-partner, the low rate data are spread using a short spreading code making his bandwidth half-occupied. While two codes with different length are dedicated to the high rate user, the first for the spreading on the user's bandwidth and the second for the spreading on the remaining part of the partner's, low rate user, bandwidth.
3.1 The Relaying Strategy
 This cooperative strategy is proposed for both the uplink and downlink transmission; the low rate user utilizes a part of his bandwidth to relay a copy of the high rate user data using a short spreading code. In this section, the procedure is explained for the uplink transmission, noting that the procedure is similarly applied for the downlink. Three OVSF spreading code lengths are required CL for the low rate user data, CH and CH′, for the high rate user data. Based on a decode-and-forward basis, the transmission is accomplished through two phases; during phase 1, the source transmits the information to the destination (base station) and the partner. We modeled the received signal at the destination and the partner in (2).
Where p is a suffix indicating the partner, and CH is the high rate user's spreading code. Meanwhile, the low rate user detects his partner data and spreads it using a short spreading code CH′. During phase 2, the low rate user spreads his data using CL on half his bandwidth and relays the high rate user data spread using CH′ on the remaining part. The transmitted signal on both users' bandwidth is shown in Figure 1 while the transmitter is shown in Figure 2a. The resulting low rate user transmitted signal follows (3).
 Where CL is the low rate user spreading code, CH′ indicates the high rate user's spreading code used on the low rate user's bandwidth. The destination decorrelates the information received from both phases and coherently combines them using a maximal ratio combiner MRC. The proposed receiver for the high rate user is illustrated in Figure 2b and both users' received data are shown in (4).
 Due to the use of OVSF orthogonal codes, the multiple access interference MAI is prevented and the users' extracted data reduce to (5).
 Two factors are carefully watched: the selection of spreading code length and the transmitting power. The selection of the spreading code length is related to the available bandwidth and the transmitted rate, noting that the code used in relaying is shorter. The total transmitted power/frame PH = PL = P is invariant for all users; the low rate user's power is equally used during the cooperative intervals to send his data and to relay the high rate user's data. However, the energy-to-noise ratio is carefully verified. The relaying strategy supports the high rate user by a wider bandwidth and higher transmitting power. An enhanced performance is expected compared to the rate-matching transmission. On the other hand, a slight degradation on the low rate user's performance is foreseen due to both the used narrower bandwidth and the used reduced transmitting power. The performance is intensively discussed in the next section.
3.2 The Bandwidth Allocation Strategy
 This strategy requires half-duplex transmission in the downlink and full duplex in the uplink transmission. This section describes the half-duplex downlink transmission; the uplink transmission is similar with the full-duplex constraint. The high rate user's data are divided using the decimation into two signals by the ratio 1/b and a/b, respectively. The part having rate a/b is spread and sent on the user's own bandwidth to the destination as in (6).
 Where XH(a/b) denotes the signal with rate a/b. While high rate user's signal with rate 1/b is spread and send on a part of the low rate user's bandwidth side by side with his own signal. The transmitted signal on both users' bandwidth is shown in Figure 3 while the transmitter is shown in Figure 4a. The resulting low rate user's transmitted signal is illustrated in (7).
 The spreading code lengths differ among both transmitted signals in (6) and (7) according to the bandwidth size and the transmitted rate. The power constraint used in the relaying strategy applies for the re-allocation strategy. The base station gathers up these signals profiting from the orthogonal spreading codes and sends it as a multiple-access signal. Each user decorrelates the received signal using his code/s; the decorrelation process of the low rate data is straightforward while the high rate user's data are expressed by (8).
 The receiver of the low rate user is a one branch matched filter while the high rate user's receiver, illustrated in Figure 4b, needs two parallel branch matched filter. One branch matched to CH to decorrelate the data from the user's signal and the other to CH′ to decorrelate the data from the partner's signal. The data received over the user's bandwidth are first processed then followed by the data received over the partner's bandwidth. The re-allocation strategy is able to support higher rates using the same resources compared to the rate-matching algorithm as will be discussed in the next section.
4 The Simulation Results
 In this section, we illustrate a CDMA-based multirate framework; the rate-matching algorithm as well as the proposed strategies is applied and the system performance is observed. The considered framework is a cellular system supporting multirate transmissions; the orthogonal variable length spreading codes OVSF are used, similar bandwidth size is assigned to all users, and the same power resources. The simulations apply Rayleigh flat fading channel; the channel coefficient remains invariant for two successive transmission intervals. However, two successive intervals form one whole cooperative interval. This assumption is adopted in most of the literature [Aazhang et al., 2003a, 2003b]. A Rayleigh channel coefficient is generated by running a Rayleigh channel and randomly selected one; the operation is repeated for every new coefficient. Based on two users' case and a Rayleigh flat fading channel, the rate-matching algorithm is first applied. The low rate user transmits rate R and uses a spreading code length of 128 while the high rate user transmits rate 4R and uses a spreading length of 32. Then the relaying and bandwidth re-allocation protocols are applied to the down link for the same transmitting rates. The low rate user utilizes a spreading code length of 64 and the high rate user is given a code length of 16 during the relaying procedure. While for the re-allocation strategy, the decimation factor is one half and two spreading codes of lengths 64 and 32 are used by the high rate user. The codes are, respectively, used for spreading on the user's bandwidth and on the partner's bandwidth. The low rate user's spreading code is of length 64.
 Under the effect of Rayleigh flat fading channel and AWGN, the probability of error is first observed for both users by applying a Rayleigh channel coefficient of 0.05. Figure 5a illustrates a slight, yet acceptable, degradation in the low rate user performance using both proposed strategies compared to the rate-matching algorithm. At low values of signal-to-noise ratio (SNR), a range of 0.5 dB performance degradation is observed while for high values of SNR nearly 1 dB degradation takes place using both proposed strategies. On the other hand, Figure 5b demonstrates that the relaying strategy supports the high rate user by 1 dB performance enhancement compared to 0.2 dB by the re-allocation strategy for high SNR range. Yet both strategies improve the high rate user performance compared to the rate-matching algorithm. Then the robustness of the proposed strategies to channel variations is investigated. The low rate user channel is kept invariant and severe channel deterioration is assumed for the high rate user by applying a Rayleigh channel coefficient of 0.01. Figure 5b illustrates that the high rate user performance using both proposed strategies surpasses the rate-matching algorithm. The relaying strategy provides the best performance profiting from the relaying transmission. It leads to 2 dB performance enhancement compared to 0.5 dB using the re-allocation strategy for high range of SNR.
 Next, the probability of outage of both users is investigated for different levels of Rayleigh flat fading channel under the constraint of the quality of service. The probability of error is restricted to a prespecified threshold of 10−3 for the low rate user and 10−6 for the high rate user. Figure 6 illustrates that the high rate user's outage is enhanced by 50% using the relaying strategy and 30% utilizing the re-allocation strategy compared to the rate-matching algorithm. While the low rate user outage is held back by only 10%. We can conclude that at the same transmission rate the relaying strategy results in better performance than the re-allocation strategy. In addition, the relaying strategy is more convenient for the uplink due to its duplexing requirement.
 Finally, the capability of the re-allocation strategy to support the higher transmission rates compared to the rate-matching algorithm is investigated using the same network resources and shorter spreading codes. Figure 7a shows a slight performance degradation on the low rate user performance when his partner's transmitted data rate is increased by 1R compared to the rate matching. While Figure 7b shows that the high rate user attains the same performance level at a higher transmitting rate 3R using the re-allocation strategy compared to 2R using the rate-matching algorithm at no additional cost. Similar results are also observed for 5R transmitting rate using the re-allocation strategy compared to 4R using the rate-matching algorithm. The bandwidth re-allocation enhances the network capability to support higher rates with no additional cost. The discussed results show remarkable performance enhancements using the proposed strategies and illustrate resistivity to channel variations due to the cooperation procedure.
 In this paper, an efficient resource allocation strategy for multirate CDMA transmissions through users' cooperation is presented. Two cooperative strategies were proposed: the relaying strategy and the bandwidth re-allocation strategy. The relaying strategy requires half-duplex for either downlink or uplink transmission on the contrary to the re-allocation strategy that requires full-duplex for the uplink and half-duplex for the downlink transmission. According to the relaying strategy procedure, the terminal with low rate requirement cooperated with those with high rate requirement by relaying a copy of their data to the destination on his own bandwidth. While for the bandwidth re-allocation strategy, the terminal with low rate requirements cooperates by sending a part of the high rate data. Either strategy provided an enhanced performance for the high rate user performance compared to the rate-matching algorithm and resulted in an acceptable performance lag for the low rate user. On the other hand, the proposed cooperative procedure was more robust to channel degradation. Moreover, the bandwidth re-allocation strategy increased the system capability to support higher transmission rates with no additional cost.
 I would like to express my deep gratitude to Prof. Said El-Khamy for his continuous support and valuable advising that resulted in our fruitful research.