Downlink Radio Resource Management Through CoMP and Carrier Aggregation for LTE-Advanced Network

Abstract

Long Term Evolution-Advanced (LTE-A) offers several new technologies to improve the performance of the user. However, poor received signal and interference from adjacent cells in the cell-edge area can reduce the efficiency of using individual technology. Therefore, the cell-edge users have lower throughput compared to the other users in the cell and LTE-A standard. An efficient downlink radio resource management scheme is proposed in this paper by combining the coordinated multipoint transmission and reception technique along with carrier aggregation technique to achieve higher throughput for the cell-edge user and better overall performance. The proposed method jointly transmits multiple component carriers to the cell-edge user from different cells to increase the bandwidth, strengthen the received signal, and reduce the interference while it satisfies several constraints. Modified largest weighted delay first packet scheduling algorithm is deployed for resource allocation, which takes into account the delay parameters, the probability of packet loss, and data rates of the user. The obtained system-level simulation results show that the proposed method significantly enhances the throughput performances, spectral efficiency, and fairness index, compared with the existing conventional methods.

Appendix

The Markov chain model has been used by previous researchers [49, 50] in the LTE system to model the correlated errors in different ways. The Markov model has two states- success and failure, as in Fig. 15. In success state, the transmitter can send a packet successfully, whereas packet transmission can be failed in a failure state. The probability of packet loss δ can be calculated by considering maximum delay, probability of transforming from one state to another state, packet arrival rate, and error probability [43].

Fig. 15

Two-state Markov chain model

For simplicity, notation of user i will not include in the following derivation. The δ can be written as

$$ \delta_{{}} = \frac{{\pi_{\tau ,1} }}{\beta } $$

(19)

Here π_τ,1 is the steady-state probability of (τ, 1) state where packets can be dropped for the extensive delay and β is the arrival rate. π_τ,1 can be expressed as follows.

$$ \pi_{\tau ,1} = \frac{1}{{C_{sum} }} $$

(20)

where

$$ C_{sum} \approx \frac{{Wx^{\tau } }}{\lambda \beta } + \frac{{W(1 - x^{\tau } )}}{\psi (1 - x)} + \frac{1}{\beta } $$

(21)

$$ C_{sum} \approx \frac{{x^{\tau } W^{2} (\beta - 1) + \lambda (\lambda \beta - W + 2\beta W)}}{\lambda \beta (\lambda \beta - W + \beta W)} $$

(22)

Here C_sum is the sum of coefficient, λ is the probability from success state to the failure state, $\Psi \approx 1- {{\upbeta }}$ when λ and β are small. So, from Eqs. (19), (20) and (22) we can write

$$ \delta = \frac{{\pi_{\tau ,1} }}{\beta } \approx \frac{\lambda (\lambda \beta - W + \beta W)}{{x^{\tau } W^{2} (\beta - 1) + \lambda (\lambda \beta - W + 2\beta W)}} $$

(23)

Success state: The error probability ε in success state with small error probability scenario is small to make the system usable. Since $\varepsilon = \lambda /\left( {\lambda + W} \right)$, W is relatively large compare to λ. Thus $\lambda /W$ is small and $x \approx 1/\left( {1 - W} \right)$. Therefore, δ can be written as follows.

$$ \delta \approx \frac{\lambda (\lambda /W\;\beta - 1 + \beta )}{{x^{\tau } W(\beta - 1) + \lambda /W(\lambda \beta - W + 2\beta W)}} $$

(24)

$$ \delta \approx \frac{\lambda (\beta - 1)}{{x^{\tau } W(\beta - 1)}} $$

(25)

$$ \delta \approx \frac{{\lambda (1 - W)^{\tau } }}{W} $$

(26)

Since $\varepsilon \approx \lambda /W$, finally, we, have

$$ \delta = \varepsilon \left( {1 - W} \right)^{\tau } $$

(27)

Failure state: In the failure state, ε is relatively large, which can make the system unusable for undue error. Since ${{\upvarepsilon = \uplambda }}/ ( {{\uplambda + W)}}$ is large, W is relatively small, thus $\lambda \gg W$. Therefore, we can write

$$ \varepsilon = {\raise0.7ex\hbox{$\lambda $} \!\mathord{\left/ {\vphantom {\lambda {(\lambda + W)}}}\right.\kern-0pt} \!\lower0.7ex\hbox{${(\lambda + W)}$}} \approx {\raise0.7ex\hbox{$\lambda $} \!\mathord{\left/ {\vphantom {\lambda \lambda }}\right.\kern-0pt} \!\lower0.7ex\hbox{$\lambda $}} = 1 $$

(28)

The terms βW, x^τW²(β − 1) and 2βW from the above Eq. (23) can be neglected as W is relatively small. We can calculate the δ for the failure state with the probability of large error as follows.

$$ \delta \approx \frac{\lambda (\lambda \beta - W + \beta W)}{{x^{\tau } W^{2} (\beta - 1) + \lambda (\lambda \beta - W + 2\beta W)}} \approx 1 \approx \varepsilon $$

(29)

Markov model is used to calculate the probability of packet loss in this paper to estimate the variable α_i and eventually gets the weighted transmission rate of the user from Eq. (5). Higher packet loss probability with shorter maximum delay makes higher α_i value and higher weighted transmission rate value. Therefore, there will be a higher chance for that user to get the RB from the scheduler. Moreover, the Markov model is comparatively less complicated and the parameters of this method match with the parameter of the MLWDF scheduling algorithm, which is used in this paper. Thus, the Markov model is a suitable approach to implement the proposed method.