Adaptive NOMA/OMA for wireless communications

Abstract

In this paper, we propose an adaptive NOMA OMA protocol that uses the best protocol between OMA and NOMA in order to maximize the total instantaneous throughput. Adaptive NOMA OMA will allow up to 1.8 dB gain compared to that of NOMA. Furthermore, adaptive NOMA OMA will offer up to 2.7 dB gain with respect to OMA. We also consider a hybrid NOMA/OMA transmission technique where some users use NOMA, and the rest of users are using OMA. We use NOMA to transmit data to K users, while OMA is employed to transmit data to the remaining $N-K$ users. The optimal value of K is determined as well as the set of NOMA and OMA users in order to maximize the total instantaneous throughput. For a throughput of 8 bit/s/Hz, adaptive NOMA OMA hybrid will offer 10 dB gain with respect to NOMA.

Appendix A

The cumulative distribution function (CDF) of the equivalent SINR at i-th NOMA user is given by

$$\begin{aligned} F(x)= & {} 1-P\left( \varGamma _{\mathrm{NOMA}}^{(N)}>x,\varGamma _{\mathrm{NOMA}}^{(N-1)}>x,\ldots ,\varGamma _{\mathrm{NOMA}}^{(i)}>x\right) \nonumber \\= & {} P\left( |h_{\mathrm{BSU}_i}|^2>\underset{i\le q \le N}{\mathrm{max}} \frac{N_0x}{E_{\mathrm{BS}}p_q-E_{\mathrm{BS}}\sum _{j=1}^{q-1}p_j}\right) \nonumber \\= & {} 1-e^{- \underset{i\le q \le N}{\mathrm{max}} \frac{N_0x}{E_{\mathrm{BS}}p_q-E_{\mathrm{BS}}\sum _{j=1}^{q-1}p_j}} \end{aligned}$$

(26)

By a derivative, we deduce the PDF of the equivalent SINR at i-th NOMA user

$$\begin{aligned} f(x)= & {} \underset{i\le q \le N}{\mathrm{max}} \frac{N_0}{E_{\mathrm{BS}}p_q-E_{\mathrm{BS}}\sum _{j=1}^{q-1}p_j}\nonumber \\&\times e^{- \underset{i\le q \le N}{\mathrm{max}} \frac{N_0x}{E_{\mathrm{BS}}p_q-E_{\mathrm{BS}}\sum _{j=1}^{q-1}p_j}} \end{aligned}$$

(27)

Therefore, $f(\gamma _{\mathrm{NOMA}}(i))$ given in (19) and (22) is equal to

$$\begin{aligned} f(\gamma _{\mathrm{NOMA}}(i))= & {} \underset{i\le q \le N}{\mathrm{max}} \frac{N_0}{E_{\mathrm{BS}}p_q-E_{\mathrm{BS}}\sum _{j=1}^{q-1}p_j}\nonumber \\&\times e^{- \underset{i\le q \le N}{\mathrm{max}} \frac{N_0\gamma _{\mathrm{NOMA}}(i)}{E_{\mathrm{BS}}p_q-E_{\mathrm{BS}}\sum _{j=1}^{q-1}p_j}} \end{aligned}$$

(28)