A New Solution for D2D Assisted 5G AR Communications


 Device-to-Device (D2D) communications are one of the main drivers of new wireless standards. D2D improves resource utilization, spectral efficiency, energy efficiency and cellular coverage of wireless networks. In some applications like online gaming, video streaming and multimedia downloading, performances can be deteriorated for users at cell edges. This is particulary emphasises in the case of augmented reality (AR) and virtual reality (VR) technologies. AR is a highly visual, interactive method of presenting relevant digital information in the context of the physical environment, e.g. connecting employees and improving business outcomes. In this paper, a new communication scheme with D2D assistance is proposed, which can significantly increase the spectral/energy efficiency in 5G AR use cases. The proposed scheme combines multiple input multiple output (MIMO) techniques, relaying and spatial modulation (SM). It allows the formation of virtual MIMO unicast, SM multicast, and SM unicast channels between AR devices. The simulation results show that spectral/energy efficiency can be significantly increased without significantly impairing bit error rate (BER) performance.


Introduction
It is expected that over 6 billion people and over 50 billion devices will be connected in the near future. That's essentially everyone and everything connected, across world, supporting every application from consumer ultrabroadband, mobile gaming, augmented reality (AR), virtual reality (VR), and autonomous connected vehicles to global business networks, ships, planes, and agriculture connected farms. Supporting such connectivity is very challenging for many current communications systems. This fact is driving the development of the new fifth generation (5G) networking architecture. The 5G networks bring ultra-broadband speeds, exponentially higher spectral and energy efficiencies, massive scalability, significantly lower costs for mobile and fixed networks, and ultra-low latency applications such as connected vehicle, device-to-device (D2D), massive machine-to-machine (M2M) and internet of things (IoT) applications. With the development of LTE-pro and 5G standards, D2D as subsystems are being developed in parallel. In a D2D wireless communication, two nearby users can communicate antennas than any of the nodes in the distributed scheme itself. Virtual antenna technology can nicely deal with the problem of lacking multiple antennas of terminals.
In this paper, we considered scenarios where there is at least one AR NN user with direct link to FN. A new signal processing at a RN is proposed, which is amplified and forward (AF) in a manner, to avoid decoding. This is especially important in terms of energy savings at RN, because decode and forward (DF) process can significantly complicate processing, especially for higher order modulations. Limited battery capacity of mobile devices is a barrier to fully exploiting the benefits of D2D communication. Meanwhile, high data rate D2D communication is required to support the increasing traffic demand of emerging applications.
The proposed scheme creates 4  X virtual MIMO/quasi orthogonal space time block code (QOSTBC) unicast channel between SN and FN. Also, new proposed processing aims to emulate distributed full orthogonal STBC (OSTBC), but the code rate remains the same as for the classical QOSTBC. SM multicast channels are formed between SN and NNs, while SM unicast channel between RN and FN is also formed.
Here, a detailed description of the distributed coding procedure for the proposed scheme is presented. The simulation results show that the proposed scheme provides a possible trade-off between bit error rate (BER) and spectral/energy efficiency. This paper is composed in the following order. In Sect. 2 the proposed scheme is described, while the simulation results are presented in Sect. 3. Conclusions are drawn in Sect. 4.

The proposed AR transmission scheme with D2D assistance
The AR D2D assisted transmission scenario is presented in Figure 1  In the first hop, multicast streams are indirectly transmitted i.e. additional spectral/time/energy resources are not occupied, while in the second hop unicast stream originated from RN is also transmitted with "zero cost" in terms of available spectral/time/energy resources.
There is a cluster of nodes, e.g. cluster that acts in the form of an AR experience, which in itself entails the need for higher data rates and lower delays. Also an important parameter is the complexity of data processing on the devices themselves as well as energy efficiency. It is always desirable to have a longer battery life.
SN uses SM technique, and there are T available transmitting antennas. Bits from multicast stream are used for determining which antennas will be active at the SN. It is assumed that two antennas are active at each time slot.
Those bits that determine the active antennas are not transmitted directly, but are decoded on the destination based on detecting which antennas are, from the possible set of antennas, used during transmission.
Unicast bits are modulated according to a certain modulation scheme and then based on the new proposed schemes are transmitted from the selected SN antennas, in order to create virtual MIMO channels between SN and FN.
In the proposed scheme, two SN antennas are active at each time slot, which means that number of possible combinations depends on the total number of antennas. If SN is equipped with only two antennas, it means that in addition to virtual MIMO, SM isn't implemented, i.e. only unicast bits are transmitted. If the number of antennas is 3, it means that three pairs of two active antennas are possible, i.e. 12 TT − , 13 TT − and 23 TT − . Three antenna combinations require at least two bits to select active antenna pairs. In order to cover all bits combinations it was proposed that one pair ( 12 TT − ) antennas has two possible transmit power states 1 P and 2 P . Table 1 shows the mapping of multicast bits for antenna selection, and combinations of active antennas and transmit powers. Table 1. One possible mapping of selecting (multicast) bits and combinations of active antennas and transmit powers,

T2-T3 P1
If the number of antennas is 4, it means that six different pairs of two active antennas are possible. Three antenna combinations require at least three bits to select antenna pairs. In order to cover all combinations with three bits, it was proposed that two antenna pairs have two possible emission states. Table 2 shows the mapping of selecting bits and combinations of active antennas and emission powers for T=4, while table 3 shows the mapping of selecting bits and combinations of active antennas and emission powers for T=5. The assumption is that channel remains the same during the four time slots. Therefore, for every four unicast STBC symbols, i.e. four time slots, depending on the number of antennas at the SN, a part of multicast bits is used for selecting active antennas.
Transmitted symbol sequences from two, i-th and j-th, selected SN antennas are: For every four time slots, a part of RN unicast bits (unicast stream 2) is used for selecting active antennas. That procedure is analog to mapping at SN.
The received signal matrices at k-th and l-th RN active antenna at four time slots are given with: In order to reduce the processing complexity at RN as well as the energy consumption new processing is performed in AF manner. There is no STBC decoding at RN. For higher order modulations this can be very time consuming.
The proposed signal processing at the RN allows distributed STBC with low complexity, so that the RN is essentially transparent for FN in terms of unicast stream from SN. Deciding comes down to looking for the minimum of the next set:

RN
Here, The received signal at r-th FN antenna is given with: Based on the proposed distributed scheme a virtual 4  R MIMO channel is created between SN and FN, Figure 2.
Decoding of indirectly transmitted bits, i.e. unicast stream originated from RN, is performed by demapping active RN antennas indexes k and l, and bits combinations.
Looking comprehensively, virtual MIMO unicast channel is formed between SN and FN, another SM unicast channel between RN and FN, and SM channels between SN and NNs, Figure 3.

Simulation results
In this section simulation results are presented for the  In addition to the obvious energy gains, it should be noted that spectral efficiency is increased, e.g. for 25%, 37.5%, and 50% at the each hop. In the case of the higher order modulation schemes, the increase in spectral efficiency is reduced because there are more directly transmitted bits within one time slot. In the case of 32QAM and 64QAM modulation, the corresponding increasing of spectral efficiency at each hop is 10%, 15%, 20% and 8.3%, 12.5%, 16.6%, respectively.
Obviously, better BER values are achieved with a system without SM, however, it can be seen from the figure that for certain considered parameters in terms of the number of antennas at SN and FN degradation of BER performance is not significant. That is, for AR applications where the key factors are higher data rate, lower latency and higher spectral and energy efficiency, a trade-off can be found in relation to the error probability as a lower priority parameter. Increasing of spectral efficiency in relation to the number of SN/RN antennas and modulation orders is given in Figure 5. As it has already been said, the decoding process is not performed at RN, but new AF-style scheme is proposed, which reduces the complexity of data processing and energy consumption, especially for higher-order modulations.
Also, virtual 4  X MIMO channels are formed between SN and FN, while a maximum of two active transmitting antennas per node are used in the transmission at each time slot. The number of receiving antennas at FN as well as the modulation scheme can be arbitrary, and do not affect or complicate the proposed transmitting procedure.
Particularly interesting deployments may be if the frequencies between SN and RN are reused between RN and FN, or if unlicensed/unauthorized bands are used.

Conclusions
In this paper, the new 5G AR transmission scheme with D2D assistance is proposed. In the considered use case it is assumed that source node has unicast stream for far node, and multicast stream for surrounding near users. Relay node is used between SN and FN. Besides unicast stream from SN, RN has another unicast stream for FN. Spatial modulation is implemented at SN and RN. At the first hop SM is used for multicast streams towards surrounding nodes, while at the second hop SM is used for direct unicast stream towards FN.
The decoding process is not performed at the RN, and the signal is transmitted in the new proposed AF manner. This reduces the complexity of data processing at RN, and saves energy, especially for higher order modulation procedures. Based on proposed processing at the RN, virtual MIMO channel is formed between the SN and FN. In terms of SM at SN/RN, only two antennas are active in each time slot, and the total number of SN/RN antennas can be arbitrary, as only two RF chains are necessary.
Simulation results for BER values are presented, and it can be concluded that there is a possible trade-off between deteriorating BER values and increased spectral/energy efficiency obtained with the proposed scheme.

Declarations
Funding -Not applicable. Without funding.

Conflicts of interest/Competing interests -Not applicable. Without conflicts of interest/competing interests.
Availability of data and material -Not applicable.
Code availability -Not applicable.