From the critical restrictions of the conventional approach mentioned in Section 3.2, this section presents the proposed NOMA-enhanced BackCom Systems. Notably, the improvement schemes for static NOMA-enhanced BackCom Systems, including two- node pairing (TNP) as well as adaptive power reflection coefficient (APRC) schemes, are first introduced. Then, the pairing schemes for dynamic NOMA-enhanced Back- Com Systems, i.e., dynamic-sized pairing (DSP) and APRC-assisted DSP schemes, are presented.
3.3.1. NOMA-Enhanced BackCom: Static Systems 3.3.1.1. Two-Node Pairing (TNP) Scheme
Instead of randomly selecting BNs for NOMA groups, the TNP scheme chooses NOMA groups based on the possibility of successfully decoding. In other words, it is expected to predict and prevent unsuccessful transmissions from NOMA groups.
Specifically, following the training phase reported in [8], B BNs are partitioned into near and far sub-regions2. Here, B/2 BNs with higher backscattered power levels, i.e., P1 ≥ P2 ≥ ã ã ã ≥ PB/2, belong to the near sub-region, while the rest of ones with PB/2+1 ≥ PB/2+2 ≥ ã ã ã ≥ PB are in the far sub-region.
The BN with the lowest power level from the near sub-region is, then, paired with the one having the lowest power level from the far sub-region. For each pair of BNs, the reader pre-estimates the SINR as in (3.2) to examine the pairing possibility. More specifically, if the SINR < γth, the selected BN in the near sub-region is controlled to transmit in the mini-slots for single-node as depicted in Fig. 3.1(b). Another BN belonging to the near sub-region with a stronger power level compared to the previous one is chosen to pair with the selected BN in the far sub-region. Otherwise, if the SINR ≥ γth, a NOMA group is created. This pairing process is performed until all BNs are considered. Finally, the reader request NOMA groups and single BNs to transmit in the timeslot containing multiple mini-slots as illustrated in Fig. 3.1(b).
In summary, the pseudo-code, which describes the reader operation of the TNP scheme, is provided in Algorithm 1. In addition, the analysis of the number of successful BN transmissions is provided in Appendix.
2Here, we mainly focus on the two-node pairing NOMA groups with M = 2, which is widely considered in the literature. It is, however, straightforward to extend for the general M case, i.e., the so-called M -node pairing scheme.
49 /* INITIATION:
for node in network do
node sets the power reflection coefficient ξ1
end /* COLLECTING:
Collect signal level of B nodes;
/* SPLITTING AND SET UP:
Sort B nodes by signal power level;
Split nodes from 1 to B/2 into Near_subregion;
Split nodes from B/2 + 1 to B into Far_subregion;
for node in Near_subregion do
node sets the power reflection coefficient ξ1
end for node in Far_subregion do
node sets the power reflection coefficient ξ2
end /* USER PAIRING SCHEME:
while Near_subregion not empty do
*/
*/
*/
*/
i ← Near_subregion.back ();
Calculate SIN Ri;;
Pair nodes i,j to a NOMA group;
Far_subregion.pop_back();
Near_subregion.pop_back();
else
Set node i to transmit in a single timeslot;
Near_subregion.pop_back();
end end
while Far_subregion not empty do j ← Near_subregion.back ();
if SINRi ⩾ γ then
Set k to transmit in a single timeslot;
end
/* TRANSMIT:
Send request NOMA groups and single nodes (if have) to transmit sequentially
k← Far_subregion.pop_back();
*/
Σγth
i
i Algorithm 1: Reader Operation for TNP Scheme.
3.3.1.2. Adaptive Power Reflection Coefficient (APRC) Scheme
A feasible solution to enhance the conventional NOMA-aided BackCom systems is the adaptive power reflection coefficient (APRC) scheme. Instead of using constant power reflection coefficients, i.e., ξi = constant, for BNs, the reader can adjust these coefficients depending on the channel conditions between the reader and BNs. This could increase the possibility of successful decoding in NOMA groups. To do so, we consider the condition for successfully decoding the i-th BN from the NOMA groups with other j-th BNs, as shown in (3.2). Then, the power reflection coefficient of ξi can be adjusted so that the i-th BN is decoded successfully, i.e.,
M j=i+1
ξ ≥ Pj + No
= ξopt, (3.3)
i Pr−2ρ i
where γth is the predefined threshold for successful decoding, while Pj is determined in (3.1). The power reflection coefficient of ξi is, then, selected in a range from ξopt to ξmax
/* INITIATION:
for i = 1 to B do BNi.ξ=ξmax ; coefficient
*/
// BNs set their power reflection
Collect signal level from B nodes;
/* SPLITTING AND SET UP:
/* COLLECTING:
Sort B nodes by signal power level;
*/
*/
K = B/M ; // number of BNs in each sub-region for i = 1 to B do
for j = 1 to B do Ri ← {};
if (i − 1) × K < j & j ⩾ i × K then end Ri.add(BNj);
end end
/* USER PAIRING SCHEME:
b = 1 ;
while K > 0 do Gm← {};
for i = 1 to B do
*/
// number of NOMA groups // NOMA group Gm.add(Ri.popRandom());
end b = b + 1;
end /* ADJUST ADAPTIVE POWER REFLECTION COEFFICIENT:
*/
for i = 1 to b do
for B = 1 to Gi.size() do K = K − 1;
end Gi[n].ξ = min(ξmax, ξoptn );
end /* TRANSMIT:
for i = 1 to b do
Send request NOMA groups Gitransmit data;
end end
*/
i
}
with ξmax ≤ 1. It is worth noting that ξi should be chosen to be the minimal value to (i) guarantee its successful decoding and (ii) reduce the interference to other BNs in the NOMA group. As a result, the coefficient ξi can be expressed as ξi = min ξmax, ξopt . In summary, the pseudo-code of reader operation is presented in Algorithm 2. In addition, the structure of BNs implementing the APRC scheme is illustrated in Fig. 3.2.
The BN is a passive node that harvests energy from an incident RF signal transmitted by the reader. The signal reflection is due to an intentional mismatch between the antenna and load impedance. By varying the load impedance, the power reflection coefficient of the i-th BN, i.e., ξi, can be adjusted. Besides, interested readers can refer to [118, Section 3.2] for more details of this structure.
Algorithm 2: Reader Operation for APRC Scheme.
51
Antenna
Figure 3.2: The structure of backscatter node with variable power reflection coefficients.
3.3.2. NOMA-Enhanced BackCom: Dynamic Systems 3.3.2.1. Dynamic-sized Pairing (DSP) Scheme
The conventional approach is applicable for static NOMA-aided BackCom systems, where the NOMA group size is fixed. This may not (i) be suitable for dynamic systems, in which the number of BNs is varied accordingly to time, and (ii) be efficient due to the fixed size of the NOMA group. To tackle these issues, the dynamic-sized pairing (DSP) scheme is presented, in which its goal is to increase the number of successful NOMA groups while supporting dynamic systems. As a result, the major differences between the DSP scheme and the conventional approach are threefold. Firstly, the DSP scheme does not virtually divide the reader’s coverage into regions, which is based on the power levels of BNs. Secondly, it is the dynamic NOMA group size, in which the number of BNs in a NOMA group is not necessarily to be M . Thirdly, the selection of BNs for NOMA groups is not random, which is similar to the TNP scheme.
In particular, the reader performs the training phase to collect information of BNs, such as the number of nodes, IDs, and backscatter powers. Also, the reflection coeffi- cients of BNs are set to be the maximum value of ξmax. The BNs are, then, sorted by the reader accordingly to their power levels, i.e., P1 ≥ P2 ≥ P3 ≥ ã ã ã . In addition, the maximum NOMA group size is determined by M , in which the number of BNs in each group is equal to or smaller than M . For NOMA grouping, the reader chooses the BN with the lowest power level to pair with BNs with higher power levels that satisfy the condition for successful NOMA decoding as in (3.2). This process is performed until all BNs are considered.
In summary, the pseudo-code of reader operation for the DSP scheme is described Backscatter signal
Incident signal Variable
impedance
c Micro- r
Switch
Information
decoder Energy
harvester ontrolle Bits
Modulation block Battery
/* INITIATION:
for i = 1 to B do */
BNi.ξ=ξmax ; // BNs set their power reflection coefficient
/* COLLECTING: */
Collect signal level of B nodes;
/* SPLITTING AND SET UP: */
Sort B nodes by signal power level;
Set BNs B = {BN1, BN2, ..., BNN};
/* USER PAIRING SCHEME:
b = 1 ;
while B.size() > 0 do Gb ← {} ;
*/
// number of NOMA groups // NOMA group
for i = M − 1 to 1 do B.top().ξ = ξM; Gb.add(B.pop());
P i threshold = γ Σ M
n=i+1 Pn + No ; for j = B.size() to 1 do
ξB[j] = ξi;
if P ⩾ P threshold
Gj m.add(BNi j); then B.remove(BNj);
break;
end end
b = b + 1;
end /* TRANSMIT:
for i = 1 to b do
Send request NOMA groups Gitransmit data;
end end end
*/
i
in Algorithm 3.
Algorithm 3: Reader Operation for DSP Scheme.
3.3.2.2. Hybrid APRC/DSP Scheme
The power reflection coefficients of BNs in the DSP scheme are, nevertheless, con- stant values. To further enhance the performance of dynamic NOMA-aided BackCom systems, the hybrid APRC/DSP scheme is introduced, which is the combination of APRC and DSP schemes. In other words, the reader can adjust the power reflection coefficients of BNs in dynamic systems using the DSP scheme. More specifically, in- stead of using the maximum power reflection coefficient for all BNs as in the DSP scheme, i.e., ξi = ξmax, these coefficients are varied accordingly to the channel con- ditions between the reader and BNs. Also, the value of ξi should be chosen as the minimum value, i.e., ξi = min ξmax, ξopt}
.