Recent Trend of Rate-Splitting Multiple Access-Assisted Integrated Sensing and Communication Systems

Abstract

In the next-generation communication systems, multiple access (MA) will play a crucial role in achieving high throughput to support future-oriented services. Recently, rate-splitting multiple access (RSMA) has received much attention from both academia and industry due to its ability to flexibly mitigate inter-user interference in a broad range of interference regimes. Further, with the growing emphasis on spectrum resource utilization, integrated sensing and communication (ISAC) technology, which improves spectrum efficiency by merging communication and radar signals, is expected to be one of the key candidate technologies for the sixth-generation (6G) wireless networks. In this paper, we first investigate the evolution of existing MA techniques and basic principles of RSMA-assisted ISAC systems. Moreover, to make the future RSMA-assisted ISAC systems, we highlight prime technologies of 6G such as non-terrestrial networks (NTN), reconfigurable intelligent surfaces (RIS), millimeter wave (mmWave) and terahertz (THz) technologies, and vehicular-to-everything (V2X), along with the main technical challenges and potential benefits to pave the way for RSMA-assisted ISAC systems.

1. Introduction

In the beyond 5G and 6G era, multiple access (MA) will play a crucial role in realizing a hyper-connected society where various intelligent machines such as robots, vehicles, and sensors are connected to the network without constraints on the data rates, coverage, transmission delay, and computing [1]. In single-input single-output (SISO) systems, orthogonal multiple access (OMA) has been mainly adopted due to its simplicity. However, since OMA merely avoids inter-user interference, its spectral efficiency performance is sub-optimal. In multiple-input multiple-output (MIMO) systems, spatial division multiple access (SDMA), which exploits abundant spatial degrees-of-freedom (DoF), is a key MA technique. Note that SDMA treats interference as noise, which can provide near-optimal spectral efficiency in weak interference regimes (i.e., underloaded systems). As the number of connected machines is explosively growing for 6G networks, practical wireless communication systems face severe interference, so SDMA is far from optimal.

In such overloaded systems, it is necessary to use a more aggressive interference management strategy such as non-orthogonal multiple access (NOMA). NOMA can enhance spectrum efficiency by simultaneously transmitting data using the same frequency and time resources. However, NOMA often encounters challenges with increased computational complexity due to successive interference cancellation (SIC) with multiple users. In addition, it is difficult to implement the optimal SIC on the receiver side, which severely degrades the NOMA performance in MIMO systems, making it even worse than SDMA.

Recently, rate-splitting multiple access (RSMA) has emerged as a key technology for flexibly managing the interference among various form-factor devices [2]. A key feature of RSMA is that it divides the user’s message into common and private parts. Private parts from all the users are separately encoded into the corresponding private message as in SDMA. On the other hand, common parts are jointly encoded into a common message. With this message construction, firstly, each user decodes and removes the common message using the SIC operation. After that, each user decodes the private message as in SDMA. By allocating power to a common message according to the dynamic interference level, RSMA can cover SDMA and NOMA as a special case. It is worth noting that RSMA empowers more efficient, robust, scalable, and flexible communication in various environments over conventional MA schemes [3]. In [4], the authors developed the RSMA power allocation to maximize the sum rate for all users with perfect channel state information (CSI). In [5], RSMA with the max-min fairness (MMF) metric was proposed for imperfect CSI conditions to maximize the minimum rate among all users.

Sixth-generation wireless networks are expected to be commercialized in the 2030s. The IMT-2030 Framework has completed the draft of 6G vision, including six usage scenarios [6]. Three of these usage scenarios are advanced versions of the enhanced mobile broadband (eMBB), massive machine-type communication (mMTC), and ultra-reliable and low-latency communication (URLLC) [7], and there are three newly introduced usage scenarios. These include (1) immersive communication, (2) massive communication, (3) hyper-reliable and low-latency communication (HRLLC), (4) ubiquitous connectivity, (5) integrated sensing and communication (ISAC), and (6) integrated AI and communication. Among these, ISAC, also referred to as joint communication and radar (JCR), radar communications (RadComs), or dual-function radar communications (DFRCs), is regarded as one of the key candidate solutions to efficiently exploit the limited spectrum resources with the explosively increasing number of connected machines.

A key feature of ISAC is to operate both sensing and communication using a shared spectrum [8]. ISAC is expected to provide not only hyper-connectivity but also robust sensing capabilities. However, a main challenge that arises from ISAC is the interference between communication and sensing due to the overlapping spectrum as well as interference between communication users, making interference management strategies crucial [9]. Recent ISAC studies have primarily focused on performing communication and radar functions on the same time and frequency resources, with the goal of maximizing spectral and hardware efficiencies [10,11]. In [12,13], novel approaches integrating MIMO and ISAC systems have been proposed. In [12], a computer vision (CV) technique has been investigated for a location-based beam focusing method in ISAC environments. Movable antennas can be exploited to enhance the performance of ISAC systems by boosting beam gain at specific angles [13].

In addition, the beamforming accuracy for the radar target is also crucial. Since RSMA can flexibly manage these interferences along with enhancing the sensing accuracy of ISAC, the sensing accuracy can be increased and communication service connectivity can be ensured in RSMA-assisted ISAC networks. In [14], the concept of RSMA-assisted ISAC was introduced, which was extended to satellite communications in [15]. Subsequently, in [16], an RSMA-assisted ISAC system incorporating a digital-to-analog converter (DAC) at the transmitter was investigated in low Earth orbit (LEO) satellite systems. In [17], the hybrid beamforming is applied in RSMA-assisted ISAC systems.

The main objective of this paper is to provide an up-to-date overview of various aspects related to RSMA-assisted ISAC systems. These include the evolution of MA techniques, key features of RSMA and ISAC, and potential research directions for RSMA-assisted ISAC systems. The main contributions of this paper are summarized as follows:

• We highlight the evolution of mobile communication with existing MA technologies from 1G to 6G. Then, we compare the key features between RSMA and conventional MA techniques such as SDMA and NOMA in terms of both message construction and resource allocation.
• We account for the basic principles of RSMA-assisted ISAC systems, along with key challenges and considerations for implementing RSMA-assisted ISAC networks.
• We put forth the potential research directions of prime candidate technologies that can boost the RSMA-assisted ISAC systems, including non-terrestrial networks (NTNs), reconfigurable intelligent surfaces (RISs), millimeter wave (mmWave) and terahertz (THz) technologies, and vehicle-to-everything (V2X).

2. Advances in Multiple Access Techniques for 6G Mobile Wireless Networks

In this section, we provide an overview of MA technologies of modern wireless communications. Then, we compare the key features of RSMA with existing MA technologies. As shown in Figure 1, since the first generation (1G) of mobile communication was introduced in the 1980s, a new generation of wireless communication systems has emerged approximately every decade. Note that the 1G system is the analog communication that supports only voice calls. In addition, since 1G was a region-specific standard, there were limitations on international roaming services. For example, the Advanced Mobile Phone System (AMPS) and C-450 were only used in the United States and Germany, respectively. As an MA technology, 1G employed frequency division multiple access (FDMA) that allocates channels to multiple users by dividing the available frequency band into non-overlapping sub-bands. The second-generation (2G) communication utilizes digital communication, enabling data transmissions such as text messages and emails. Compared to 1G, the roaming services became more widely available since many European countries employ the same standard such as Global System for Mobile Communications (GSMs). The MA technique used in 2G GSM is time division multiple access (TDMA). TDMA divides the signal into multiple time slots by sharing the same frequency band.

The third generation (3G) refers to the era of multimedia services where the Internet and video calls are available. The standard of 3G communication is based on the Universal Mobile Telecommunication System (UMTS) and the MA technology of 3G is code division multiple access (CDMA), which allows multiple users to simultaneously share the same frequency band by using different codes. It is worth noting that there was a substantial increase in mobile communication subscribers due to the introduction of smartphones with multi-touch screens around 2007.

The key feature of the fourth-generation (4G) communication is the true realization of mobile Internet, supporting services such as video streaming, gaming, and social network service (SNS). Note that the international standard leading 4G is Long-Term Evolution (LTE). Two key technologies in LTE are orthogonal frequency division multiple access (OFDMA) and MIMO.

The fifth generation (5G) was commercialized in 2019 and three key usage scenarios of 5G are eMBB, mMTC, and URLLC. To provide much higher data rates and lower latency compared to the previous generations, 5G exploits new frequency bands such as mmWave. Fifth-generation systems also adopt various innovative technologies such as massive MIMO and network slicing. Massive MIMO utilizes a large number of antennas to perform high-precision beamforming and network slicing allows for the customized networks tailored to any key usage scenarios. It is worth noting that the MA technology remains OFDMA; on the other hand, while OFDMA was only used for downlink in 4G, in 5G, it is applied to both uplink and downlink.

The MA technologies described so far are OMA, which divides frequency and time resources to ensure orthogonality. As the number of connected devices is growing explosively, it is burdensome for OMA to serve these devices simultaneously. To address this challenge, it is necessary to enable multiple users to share the same time and frequency resources.

2.1. Spatial Division Multiple Access

In wireless communication systems, achieving high spectral efficiency requires serving multiple users through effective MA technologies. In MIMO systems, the key MA technology is SDMA. From an information-theoretic perspective, SDMA can be seen as employing an interference management strategy that treats interference as noise. SDMA provides near-optimal spectral efficiency when interference between users is minimal. However, in practical wireless systems, channel feedback errors make it difficult to guarantee perfect CSI in the downlink. As a result, the performance of SDMA is often far from optimal.

2.2. Non-Orthogonal Multiple Access

Compared to SDMA, a more aggressive interference management strategy is NOMA, which decodes and removes the interference through SIC operation. NOMA can increase the number of serving users and improve spectral efficiency compared to traditional OMA technologies. NOMA can be implemented using code and power domains within the same time-frequency resources [18]. In code-domain NOMA, user-specific spreading sequences are used to share the total resources, while in power-domain NOMA, the difference in channel gains between users is utilized for multiplexing through power allocation. The main idea behind the NOMA implementation lies in using superposition coding (SC) at the transmitter to transmit data and SIC at the receiver. However, in practice, it is challenging to achieve optimal SIC order in MIMO systems. This can severely degrade the performance of NOMA, and in some cases, its performance may even be worse than that of SDMA.

2.3. Rate-Splitting Multiple Access

Existing multiple access technologies, SDMA and NOMA, handle interference by treating it as noise or fully decoding the interference, respectively [19]. Note that RSMA combines these two philosophies. The key idea of RSMA is to split each user’s message into a common part and a private part. The common parts from multiple users are combined and encoded as a common message. The private parts of each user are encoded as private messages. At the receiver, each user first decodes the common message and removes it using SIC, and then decodes their private message. When the common message rate is appropriately determined, RSMA can mitigate interference between users and achieve better spectral efficiency compared to existing MA techniques.

2.4. Comparison of RSMA Technology with Existing Multiple Access Technologies

By adjusting the ratio and power of common and private messages, RSMA can include NOMA, SDMA, OMA, and multicasting techniques as special cases. As shown in Figure 2, through a two-user message construction example, we can show that existing MA technologies can be regarded as special cases in RSMA. Recall that RSMA splits each user’s message into a common and private part. For example, user 1’s message $W_1$ is divided into a common part $W_{c,1}$ and a private part $W_{p,1}$. Similarly, user 2’s message $W_2$ is divided into a common part $W_{c,2}$ and a private part $W_{p,2}$. Here, $W_{c,1}$ and $W_{c,2}$ are combined to form a common message $W_c$, while $W_{p,1}$ and $W_{p,2}$ become the private message for each user.

In the case of SDMA, the power for the common message is set to zero in RSMA, which means that users’ messages consist of only private messages. In other words, user 1’s message $W_1$ consists entirely of the private part $W_{p,1}$, and user 2’s message $W_2$ consists entirely of the private part $W_{p,2}$. For NOMA, one user has only the common message and the other user has only private message. For example, if user 1 has a higher channel gain than user 2, then user 1’s message $W_1$ becomes entirely the common message $W_c$, while user 2’s message $W_2$ becomes entirely the private message $W_{p,2}$. In this case, user 1 completely decodes user 2’s common message and removes this interference through the SIC operation. Meanwhile, user 2 decodes its own message by regarding user 1’s private message as noise.

In OMA, since a single user occupies a wireless resource at a time, only user 1’s message $W_1$ is used entirely as the private message $W_{p,1}$ and transmitted while user 2’s message $W_2$ is not transmitted. In a different time slot or frequency, user 2’s message $W_2$ is used entirely as the private message $W_{p,2}$ and fully transmitted, while user 1’s message $W_1$ is not used. In the case of multicasting, the same message is transmitted to both users as the common message. Similar to RSMA, multicasting uses a common message that all users decode. The difference between multicasting and RSMA lies in how the common message is composed. In general, the common message in multicasting is generated based on service demands, such as video streaming. In contrast, RSMA designs the common message to mitigate interference between users.

It is worth noting that RSMA is robust to imperfect CSI compared to existing MA techniques. When perfect CSI is available at the transmitter, linear precoding techniques such as SDMA can effectively manage interference between users. However, since SDMA treats interference as noise (TIN), it can suffer from significant performance degradation under imperfect CSI conditions. SDMA achieves near-optimal performance when the inter-user interference is weak in the wireless communication systems with perfect CSI at the transmitter (CSIT). However, in general, perfect CSIT cannot be guaranteed in practical wireless communication systems due to the limited feedback resources. In such a wireless environment, TIN is far from optimal. This encourages the use of more aggressive types of interference management schemes such as a joint decoding (JD) type. As a JD type of interference management strategy, NOMA achieves near-optimal performance when one interference link is much stronger than the other. It is worth noting that RSMA can embrace both TIN and JD types of interference management strategies. Since RSMA can include both SDMA (TIN) and NOMA (JD) as special cases by adjusting the common message’s rate, power, and content, RSMA can outperform existing MA techniques in all scenarios [5].

Specifically, when the rate of the common message is zero, RSMA reduces to SDMA. Additionally, if the common message contains only a specific user’s portion and the corresponding user has no private message, then RSMA becomes NOMA. Since NOMA fully decodes the message of other users, the order of SIC implementation is crucial. An improper SIC order can significantly degrade performance, sometimes even falling below that of SDMA. Moreover, while NOMA has no flexibility in message construction, RSMA constructs a common message that includes parts of each user’s message, simplifying the SIC order to just two steps. Consequently, RSMA is more resilient than NOMA. Further, RSMA has scalability with an increasing number of users. When the number of users grows, both SDMA and NOMA require more accurate CSI and involve more complexity in the process of interference management. On the other hand, RSMA can dynamically adjust the ratio of common and private messages, leading to less performance degradation with an increasing number of users and offering excellent scalability.

3. RSMA-Assisted Integrated Sensing and Communication Systems

In this section, we first provide the RSMA system model and then investigate the RSMA-assisted ISAC architecture. Finally, we study the performance metrics of RSMA-assisted ISAC systems.

3.1. RSMA System Model

RSMA has been studied in various forms depending on how user messages are split and common messages are combined for multiple users. In this paper, we focus on the key operations of one-layer downlink RSMA (see more comprehensive studies in [20,21]).

We consider the downlink RSMA system where the base station (BS) employs M transmit antennas and K users have a single antenna. By denoting the user set $\mathcal{K}=\{1,2,\dots ,K\}$, as shown in Figure 3, the message of the k-th user ($\forall k\in \mathcal{K}$), $W_k$, is split into a common part $W_{c,k}$ and a private part $W_{p,k}$. The common parts of all users, $\{W_{c,1},W_{c,2},\dots ,W_{c,K}\}$, are combined into a single common message $W_c$. The common message $W_c$ and the K private messages $\{W_{p,1},W_{p,2},\dots ,W_{p,K}\}$ are independently encoded into $s_c$ and $\{s_1,s_2,\cdots ,s_K\}$, respectively. By defining ${\mathbf{p}}_c\in {\mathbb{C}}^{N_t\times 1}$ and ${\mathbf{p}}_k\in {\mathbb{C}}^{N_t\times 1}$ as the precoding vectors for the common and private streams, the transmitted signal $x\in {\mathbb{C}}^{N_t\times 1}$ can be expressed as

$$\mathbf{x}={\mathbf{p}}_c{s}_c+\sum _{k\in \mathcal{K}}{\mathbf{p}}_k{s}_k.$$

(1)

The overall transmitted symbol streams can be expressed as $\mathbf{s}={[s_1,s_2,\dots ,s_K,s_c]}^{\mathrm{T}}\in {\mathbb{C}}^{(K+1)\times 1}$, following $E\left[{\mathbf{ss}}^H\right]={\mathbf{I}}_{K+1}$. The transmit power constraint is expressed as $P_c+{\sum }_{k\in \mathcal{K}}{P}_k\le P$, where $P_c={\parallel {\mathbf{p}}_{\mathbf{c}}\parallel }^2$ and $P_k={\parallel {\mathbf{p}}_{\mathbf{k}}\parallel }^2$. The received signal at the k-th user can be expressed as

$$y_k={\mathbf{h}}_k^H\mathbf{x}+n_k={\mathbf{h}}_k^H{\mathbf{p}}_c{s}_c+{\mathbf{h}}_k^H\sum _{k\in \mathcal{K}}{\mathbf{p}}_k{s}_k+n_k$$

(2)

where ${\mathbf{h}}_k\in {\mathbb{C}}^{N_t\times 1}$ is the channel vector between the transmitter and the k-th user, and $n_k\sim \mathcal{CN}(0,{\sigma }_k^2)$ is additive white gaussian noise (AWGN). Note that the first step on the receiver side is applying SIC to detect the common stream $s_c$ and detect the common message part ${\widehat{W}}_{c,k}$ by treating the private streams as noise. The signal-to-interference plus noise ratio (SINR) for the common stream at the k-th user can be expressed as

$${\gamma }_{c,k}={\displaystyle \frac{|{\mathbf{h}}_k^H{\mathbf{p}}_c{|}^2}{{\sum }_{j\in \mathcal{K}}{\left|{\mathbf{h}}_k^H{\mathbf{p}}_j\right|}^2+{\sigma }_k^2}}.$$

(3)

The common stream is reconstructed using the estimated ${\widehat{W}}_c$ and subtracted through the SIC operation. Then, the private message of the k-th user ${\widehat{W}}_{p,k}$ can be detected by treating the other users’ private streams as noise. Considering the perfect SIC operation, the SINR for the private stream at the k-th user can be expressed as

$${\gamma }_{p,k}={\displaystyle \frac{|{\mathbf{h}}_k^H{\mathbf{p}}_k{|}^2}{{\sum }_{j\in \mathcal{K},j\ne k}{\left|{\mathbf{h}}_k^H{\mathbf{p}}_j\right|}^2+{\sigma }_k^2}}.$$

(4)

The maximum rate for the common stream is $R_c=min\{{log}_2(1+{\gamma }_{c,1}),\dots ,{log}_2(1+{\gamma }_{c,K})\}$, and the rate for the private stream is $R_{p,k}={log}_2(1+{\gamma }_{p,k})$. Since the common stream contains the common messages intended for all users, we have $R_c={\sum }_{k\in \mathcal{K}}{C}_k$, where $C_k$ represents the portion of the common stream rate allocated to the k-th user. Therefore, the total achievable rate for the k-th user is $R_{k,\mathrm{tot}}=C_k+R_{p,k}$.

3.2. RSMA-Assisted ISAC Architecture

In this section, we provide an overview of the interaction between RSMA and ISAC, as shown in Figure 4. Note that RSMA is applied in ISAC systems to control inter-function interference (i.e., interference between radar and communication functions) as well as intra-function interference (i.e., interference among communication users). Note that the key difference between RSMA-ISAC and RSMA is that the common stream of RSMA-ISAC serves to form a beam towards the radar target. Thus, the multi-functionality of the common stream in RSMA-ISAC enables efficient operation of joint sensing and communication even without a dedicated radar signal. In other words, RSMA-ISAC employs a dynamic interference management strategy to optimize the balance between sensing accuracy and communication reliability, making it more adaptable and robust to operational scenarios compared to RSMA. Recall that ISAC is expected to play a pivotal role in next-generation wireless networks, achieving both hyper-connectivity and reliable sensing capabilities. Note that ISAC incorporates wireless communication and remote sensing functionalities into a unified system, where both functions share the spectrum and hardware.

Note that in the monostatic ISAC systems, BS has $N_t$ transmit antennas and $N_r$ receive antennas, enabling serving K users and detecting moving targets simultaneously. For simplicity, we consider a one-layer RSMA model as described in Section 3.1. Let $\mathbf{s}\left[t\right]={[s_1\left[t\right],s_2\left[t\right],\dots ,s_K\left[t\right],s_c\left[t\right],s_R\left[t\right]]}^T\in {\mathbb{C}}^{(K+2)\times 1}$ denote the baseband signal before the precoding is applied where $t\in \mathcal{T}=\{1,2,\dots ,T\}$ denotes the discrete-time index within one coherent processing interval (CPI) and $s_R\left[t\right]$ is a radar sequence. Note that both the transmitter and receiver typically know $s_R$ in advance. If T is sufficiently large and the data streams are independent, then ${\displaystyle \frac{1}{T}}{\sum }_{t=1}^T\mathbf{s}\left[t\right]\mathbf{s}{\left[t\right]}^H={\mathbf{I}}_{K+2}$ is satisfied. The transmit signal at time index t is expressed as

$$\mathbf{x}\left[t\right]=\mathbf{P}\mathbf{s}\left[t\right]={\mathbf{p}}_R{s}_R\left[t\right]+{\mathbf{p}}_c{s}_c\left[t\right]+\sum _{k\in \mathcal{K}}{\mathbf{p}}_k{s}_k\left[t\right]$$

(5)

where the precoding matrix $\mathbf{P}=[{\mathbf{p}}_1,{\mathbf{p}}_2,\dots ,{\mathbf{p}}_K,{\mathbf{p}}_c,{\mathbf{p}}_R]\in {\mathbb{C}}^{N_t\times (K+2)}$ is maintained over one CPI. To be specific, $\{{\mathbf{p}}_1,{\mathbf{p}}_2,\dots ,{\mathbf{p}}_K\}$ are the precoding vectors for K private streams while ${\mathbf{p}}_c$ and ${\mathbf{p}}_R$ are the precoding vectors for the common stream and radar sequence, respectively. The covariance matrix of the transmit signal can be written as ${\mathbf{R}}_x={\displaystyle \frac{1}{T}}{\displaystyle \sum _{t=1}^T}\mathbf{x}\left[t\right]\mathbf{x}{\left[t\right]}^H=\mathbf{P}{\mathbf{P}}^H$. Note that the transmit power budget is $P_t$ (i.e., $\mathrm{tr}\left(\mathbf{P}{\mathbf{P}}^H\right)\le P_t$).

For the received signal in the communication model, each user receives the signal as follows

$$\begin{array}{cc}\hfill y_k\left[t\right]& ={\mathbf{h}}_k^H\mathbf{x}\left[t\right]+n_k\left[t\right]\hfill \\ \hfill & ={\mathbf{h}}_k^H{\mathbf{p}}_R{s}_R\left[t\right]+{\mathbf{h}}_k^H{\mathbf{p}}_c{s}_c\left[t\right]+{\mathbf{h}}_k^H\sum _{k\in \mathcal{K}}{\mathbf{p}}_k{s}_k\left[t\right]+n_k\left[t\right].\hfill \end{array}$$

(6)

Firstly, each user decodes the common stream by treating other streams as interference. The achievable common rate for the k-th user can be expressed as

$$R_{c,k}={log}_2\left(1+{\displaystyle \frac{|{\mathbf{h}}_k^H{\mathbf{p}}_c{|}^2}{{\sum }_{i\in \mathcal{K}}|{\mathbf{h}}_k^H{\mathbf{p}}_i{|}^2+\delta {\left|{\mathbf{h}}_k^H{\mathbf{p}}_R\right|}^2+{\sigma }_k^2}}\right)$$

(7)

where $\delta$ is the binary parameter related to the SIC operation of the radar sequence. When the SIC for $s_R$ is applied, $\delta$ becomes 0.

After the common stream is subtracted from the received signal, each user decodes the private stream. The achievable private rate for the k-th user can be expressed as

$$R_{p,k}={log}_2\left(1+{\displaystyle \frac{|{\mathbf{h}}_k^H{\mathbf{p}}_k{|}^2}{{\sum }_{i\in \mathcal{K},i\ne k}|{\mathbf{h}}_k^H{\mathbf{p}}_i{|}^2+\delta {\left|{\mathbf{h}}_k^H{\mathbf{p}}_R\right|}^2+{\sigma }_k^2}}\right).$$

(8)

Then, the total achievable rate for the k-th user is $R_{p,k}+C_k$, where $C_k$ is the k-th user’s common rate portion. Since the common stream is decoded by all users, the sum of the common rate portions ${\sum }_{k\in \mathcal{K}}{C}_k$ should be less or equal than ${min}_k{R}_{c,k}$.

For the reflected signal, BS receives the $N_r\times 1$ echo signal at time index t as follows

$$\begin{array}{cc}\hfill {\mathbf{y}}_r\left[t\right]& =\alpha e^{j2\pi {\mathcal{F}}_{D}t{T}_{\mathrm{sym}}}\mathbf{b}\left(\theta \right){\mathbf{a}}^T\left(\theta \right)\mathbf{x}\left[t\right]+{\mathbf{n}}_r\left[t\right]\hfill \end{array}$$

(9)

where $\alpha$ is the complex reflection coefficient related to the radar cross-section (RCS). ${\mathcal{F}}_D={\displaystyle \frac{2v{f}_c}{c}}$ is the Doppler frequency, where v is the velocity of the radar target. $f_c$ and c are the carrier frequency and the speed of light, respectively. $T_{\mathrm{sym}}$ denotes the symbol period. $\theta$ denotes the radar target’s direction of departure (DoD) as well as the direction of arrival (DoA) of the target, which are the same in the monostatic system. $\mathbf{a}\left(\theta \right)\in {\mathbb{C}}^{N_t\times 1}$ and $\mathbf{b}\left(\theta \right)\in {\mathbb{C}}^{N_r\times 1}$ are the transmit and receive steering vectors, respectively. ${\mathbf{n}}_r\left[t\right]$ is AWGN.

For the joint design of communication and radar functions, communication performance metrics and radar performance metrics should be jointly optimized. Specifically, the minimum fairness rate [22], energy efficiency [23], and weighted sum rate [14] can be considered as the communication performance metrics. As the radar sensing metrics, the beampattern matching [24] and Cramér–Rao bound (CRB) [25] are considered.

For the beampattern matching, the mean square error (MSE) for the beampattern can be exploited as $\mathrm{MSE}={\sum }_{m=1}^M{\left|P_d\left({\theta }_m\right)-{\mathbf{a}}^H\left({\theta }_m\right){\mathbf{R}}_{\mathbf{x}}\mathbf{a}\left({\theta }_m\right)\right|}^2$ where ${\theta }_m$ is the m-th azimuth angle grid among all M grids, $P_d\left({\theta }_m\right)$ is the desired beam pattern level at ${\theta }_m$. $\mathbf{a}\left({\theta }_m\right)={[1,e^{j2\pi \Delta sin\left({\theta }_m\right)},\dots ,e^{j2\pi (N_t-1)\Delta sin\left({\theta }_m\right)}]}^T\in {\mathbb{C}}^{N_t\times 1}$ is the transmit steering vector and $\Delta$ is the normalized distance between adjacent array elements. ${\mathbf{R}}_{\mathbf{x}}$ is the covariance matrix of transmit waveforms. A smaller MSE results in a higher signal-to-noise ratio (SNR) at the receiver, leading to improved target detection probability and estimation accuracy [14].

In addition, CRB can be exploited for target estimation. CRB is a lower bound on the variance of an unbiased estimator, which can be obtained by calculating the inverse of the Fisher information matrix (FIM), i.e., CRB = ${\mathbf{F}}^{-1}$. The matrix $\mathbf{F}$ contains the estimation of the angular direction, the complex reflection coefficient, and the Doppler frequency of the radar echo signal [26].

Since RSMA-assisted ISAC systems offer flexibility in both message construction and resource allocation, RSMA-assisted ISAC systems outperform ISAC systems based on SDMA and NOMA. In [17], RSMA-assisted ISAC systems show better communication-sensing performance trade-offs, higher energy efficiency, and robust interference management compared to both SDMA-based ISAC and NOMA-based ISAC schemes. In [27], RSMA-assisted ISAC systems demonstrate better spectral efficiency compared to NOMA-based ISAC in IoT environments where CSI is imperfect or interference cancellation is limited. Further, since RSMA can effectively manage mutual interference between communication and radar functions by dynamically allocating resources in the spatial and power domains, which outperforms ISAC systems with the orthogonal allocation of time or frequency resources [28]. In a nutshell, the performance gain of the RSMA-assisted ISAC systems mainly comes from the three crucial roles of the common stream: (1) intra-system interference management, (2) inter-system interference management, and (3) beam forming to the target.

3.3. Key Challenges for RSMA-Assisted ISAC

3.3.1. Joint Waveform Design for RSMA-Assisted ISAC

The main goal of joint waveform design is to achieve a better communication and sensing performance trade-off. One can utilize the existing communication waveforms, such as orthogonal frequency division multiplexing (OFDM) [29]. In [30], it is shown that OFDM-ISAC improves the resolution of the velocity estimation without increasing the signal bandwidth.

Due to the significant performance degradation of conventional OFDM in high-speed environments, orthogonal time frequency space (OTFS) modulation has received considerable attention to provide more reliable communication under such conditions [31]. Unlike the OFDM-ISAC system, which operates in the time-frequency domain, the OTFS-ISAC system operates in the delay-Doppler domain, providing resilience to Doppler spread and delay [32]. In [33], a new OTFS system model is introduced, utilizing ISAC to analyze channel estimation error probability and outage probability. Additionally, ref. [34] extends the OTFS-ISAC system to vehicular applications, offering a communication system capable of predicting vehicle states.

However, such communication waveforms can experience high peak-to-average power ratio (PAPR) and resolution degradation, which can degrade detection performance [35]. To address these issues, recent research has focused on integrated waveform design that considers both communication and sensing performance. In [36], a joint waveform design was proposed to minimize multi-user interference, while [37] analyzed the integrated performance trade-offs in Gaussian channels based on the CRB. Furthermore, ref. [38] performed joint optimization based on mutual information in MIMO systems.

Recently, RSMA-ISAC waveform design has been proposed by jointly minimizing CRB and maximizing MFR [22]. In addition, in [39], the RIS has been considered in the RSMA-ISAC waveform design. In [40], a transmissive RIS framework was proposed to improve secrecy energy efficiency and CRB under imperfect CSI conditions.

3.3.2. Dual-Functional Beamforming Optimization

It is important to optimize the beamforming to both maximize the communication performance such as weighted sum rate and the minimum rate among all users while ensuring the radar performance by satisfying the CRB constraints.

Extensive research has focused on optimizing ISAC beamforming. In [41], the joint optimization of full-duplex (FD) based ISAC systems has been investigated under the criterion of minimizing the transmit power. In [42], hybrid beamforming for FD-based systems is optimized. The joint design of communication reception coefficients and transmission waveforms is addressed in [43], and it is shown that the proposed technique achieves high communication quality of service (QoS). Additionally, ref. [44] introduced a finite alphabet input method to propose a block-level beamforming design for ISAC systems.

Most of the literature on ISAC beamforming addressed the interference between multiple users and radar targets via SDMA. To achieve more flexible interference management, Refs. [10,14] investigated RSMA-based ISAC systems. Since RSMA allows partial decoding of interference while treating some interference as noise [45], enabling robust interference management strategy for both inter-function and intra-function interference in future ISAC systems. Furthermore, ref. [26] extended RSMA-based ISAC to multiple target scenarios.

3.3.3. Further RSMA-Assisted ISAC Challenges and Considerations

In order to effectively enable RSMA-assisted ISAC systems, the following technical challenges must be addressed:

Energy efficiency: In RSMA-ISAC systems, there are two key factors for maximizing energy efficiency: the use of low-resolution DACs and the management of interference between communication and radar functions. Low-resolution DACs can significantly reduce power consumption and then bring lower hardware costs [46]. However, reduced resolution can lead to signal quality, making it crucial to optimize power consumption while maintaining performance. Since communication and radar functions share the same hardware, minimizing interference is critical. To control the inter-function and intra-function interference effectively while also enhancing energy efficiency, advanced interference management technique such as RSMA can be applied.

Compatibility and interoperability: Ensuring compatibility and interoperability between radar and communication functions in RSMA-ISAC systems is crucial for seamless integration. Compared to the existing MA technique such as SDMA, RSMA can provide a more flexible solution by using the common stream to manage interference between radar and communication simultaneously [14]. RSMA-based ISAC systems can ensure high-quality communication performance and improved spectrum utilization, enhancing the interoperability and efficiency of the system across various deployment scenarios.

Security and data protection: Note that the security design in RSMA-assisted ISAC networks faces considerable challenges. Although RSMA manages the interference by properly dividing the data stream into common and private streams, interference management is particularly critical; while RSMA manages interference by dividing data into common and private streams, controlling interference to prevent unauthorized eavesdropping is quite challenging [47]. Moreover, imperfect CSI plays a vital role in security. Since the precise beamforming relies on accurate channel information, channel estimation errors can increase the risk of interception by unauthorized users. Note that these challenges are major obstacles in simultaneously ensuring security and performance in RSMA-assisted ISAC networks, necessitating the development of advanced algorithms and technologies to overcome these challenges [40].

Channel estimation: Recently, integrating MIMO and ISAC systems has emerged as a promising solution for enhancing both communication and sensing capabilities in ISAC environments [12,13]. In [12], a CV-based beam focusing technique has been investigated to accurately estimate the position of UE and adjust the beam accordingly, reducing the delays and overhead associated with conventional beam sweeping. In [13], movable antennas are exploited to improve ISAC system performance. Exploiting movable antennas can optimize beamforming gain at target locations by boosting beam gain at specific angles. However, as the position of movable antennas changes, the variation of the channel information also increases, making channel estimation more complex than in fixed antenna systems. To this end, tensor decomposition-based channel estimation can be applied to provide precise channel information with fewer pilots [13]. This enables ISAC systems with movable antennas to achieve better transmission rates and beamforming gains compared to fixed antenna systems.

4. Potential Research Directions of RSMA-Assisted ISAC Systems

In this section, we explore the potential research directions of RSMA-assisted ISAC with other emerging communication technologies. In

Table 1. Four prime technologies for RSMA-assisted ISAC.

Prime Technology	Key Features	Future Application Scenarios
NTN	– Improved resource utilization [48] – Reduction in radar echo path loss [49] – Provides wide coverage [50] – Improved detection accuracy [51]	– Efficient management and detection of beam interference between communication users [48] – Real-time communication and sensing capabilities in disaster response systems [52]
RIS	– Improved energy efficiency [53] – Improved communication performance [54] – Iterative optimization for SNR maximization [39]	– Enable communication and localization in smart cities [55] – Intelligent control of wireless channels at a low cost [39]
MmWave and THz	– Pilot overhead reduction via compressed sensing [56] – Boosts reliability via beamforming [17]	– Enhanced communication and detection accuracy [57]
V2X	– Higher spectral efficiency [58] – High-precision traffic environment perception [59]	– Provision of high-quality communication services through on-board units [59]

, we highlight four prime technologies, namely, NTN, RIS, mmWave and THz technologies, and V2X, as represented in Table 1.

4.1. Non-Terrestrial Networks

Since terrestrial networks (TNs) are not deployed in remote areas, such as deserts and oceans, NTNs can effectively provide high-speed communication services to such areas. NTN terminals are classified into satellites, high-altitude platform stations (HAPSs), and unmanned aerial vehicles (UAVs) based on their flight altitude. Satellite terminals are further divided into geostationary orbit (GEO), medium earth orbit (MEO), and LEO based on their orbital altitude. Recently, LEO satellites have attracted interest from both academia and industry because of their high transmission rates and flexibility. In recent years, the combination of LEO and ISAC has increased throughput by managing both sensing and communication [11].

Note that since LEO satellites offer significantly reduced latency and path loss compared to GEO and MEO satellites, they have led to explosive research in this area [60]. However, LEO satellites, due to their higher mobility compared to other satellites or TNs, face challenges such as Doppler shifts and imperfect CSI. Among the existing multiple access techniques, RSMA can be an effective means of satisfying users’ QoS even under imperfect CSI conditions, and it has been shown to outperform other MA technologies in satellite networks [61].

Most of the literature on RSMA has been focused on TNs. In [48], RSMA was applied in LEO-ISAC systems to efficiently manage and detect beam interference among communication users in LEO-ISAC systems. Figure 5 illustrates an overview of the RSMA-assisted LEO-ISAC systems. Note that since a high round-trip propagation delay results in a severe echo path loss in a monostatic ISAC structure, a bistatic ISAC can be a promising scenario where a radar receiver is separated from the transmitter. In [49], bistatic RSMA-assisted LEO-ISAC systems have been investigated to reduce radar echo path loss.

Further, with recent technological advances, UAVs are being widely used in various applications such as communication, search and rescue, and remote monitoring. Most of the literature on UAV-assisted ISAC research has focused on channel modeling and performance, utilizing ISAC to support both communication and sensing capabilities [50]. In [51], the collaborative networking system for an intelligent UAV cluster was investigated using UAVs and ISAC to improve sensing accuracy. In [62], the resource allocation through trajectory or deployment optimization was investigated in UAV-assisted ISAC systems. Recent studies [63] have demonstrated that RSMA in UAVs outperforms NOMA in terms of energy efficiency. In [64], the initial framework and maximized energy efficiency in a radar detection and communication system were proposed. Recently, ref. [52] introduced an RSMA-assisted ISAC system model in emergency UAV systems, demonstrating higher improved weighted sum rate (WSR) performance.

4.2. Reconfigurable Intelligent Surface

RIS has received considerable attention in next-generation wireless communication systems, due to their ability to actively manipulate the radio environment and effectively reduce interference between multiple users [65]. In recent years, ISAC systems that exploit RIS have been extensively studied [66]. In such systems, beamforming technology can be effectively adopted to improve target detection and parameter estimation performance [67]. In [68], spare arrays with partially active elements are exploited for channel estimation in RIS-assisted ISAC systems. Recently, RSMA has been further considered in RIS-assisted ISAC systems [39], aiming to maximize the SNR for target detection. In [39], a study on RSMA-RIS-assisted ISAC systems was conducted for the first time, maximizing the SNR for target detection. The overview diagram of the RSMA-assisted RIS-ISAC systems is shown in Figure 6.

4.3. Millimeter Wave and Terahertz Technologies

MmWave and THz bands have been considered as core spectrum for 6G wireless networks. To fully leverage the abundant frequency resource in mmWave and THz bands, beamforming techniques based on massive MIMO antenna arrays have been popularly used to compensate for the severe path loss in higher frequency bands. One viable technique using these bands is cell-free massive MIMO (CF-mMIMO) systems that can significantly improve the data transmission range and performance by deploying access points (APs) distributed in a wide area to collectively serve user equipment (UE) [69]. The combination of RSMA and CF-mMIMO provides a broad coverage area via CF-mMIMO and enhances interference management through RSMA, leading to improved energy efficiency [70]. This can address the high propagation loss in mmWave and THz communication and the cost issues associated with many RF chains.

In addition, due to the need for wider bandwidth, research on hybrid ISAC in the 30–300 GHz spectrum is being investigated to enhance both communication performance and sensing accuracy [57]. In [71], the use of wider bandwidth in THz bands enables high-resolution sensing. Thus, THz-based ISAC systems can enhance data transmission rate and high-resolution sensing capabilities, improving both spectral and energy efficiency. However, due to the high directivity and substantial atmospheric absorption in the THz band, an advanced interference management technique based on highly directional and adaptive beamforming is required to perform the sensing in the THz band. Note that the synergistic design of THz bands and ISAC systems can be applied in various fields such as smart cities and intelligent factories [72]. Recently, the integration of radar and massive MIMO communication has been studied to fully exploit vast spatial degrees of freedom and abundant bandwidth resources in ISAC systems [73]. To this end, accurate channel training is essential for effective target sensing functionality, and CV technology can be applied to analyze sensing information, enabling precise understanding of wireless environments [74]. The advantages of RSMA in multi-antenna environments have been demonstrated for addressing multi-user interference in mmWave communication [75]. In [17], it is shown that RSMA-assisted mmWave-ISAC (see Figure 7), using hybrid beamforming algorithms, outperforms both NOMA-ISAC and SDMA-ISAC.

4.4. Vehicle-to-Everything

V2X is an emerging technology for 6G wireless networks, supporting diverse traffic scenarios [76]. As the number of connected users is growing explosively, there would be strong interference in V2X networks by sharing the network infrastructure with celluar networks [77]. Prior research on RSMA-assisted V2X networks has demonstrated that RSMA can effectively mitigate interference and enhance data transmission rates. In [58], V2X systems utilizing RSMA demonstrated superior performance at higher SNR levels, achieving an average of 25% higher spectral efficiency compared to NOMA technology. Since RSMA showed its ability to maintain high spectral efficiency even in high-mobility scenarios, this advantage is especially relevant for satellite-to-vehicle networks, as it heavily depends on the accuracy of CSI estimation. In [78], a non-cooperative game based on an incentive mechanism was proposed, showing that RSMA-based satellite-to-vehicle networks outperform other MA technologies in terms of WSR.

Recently, there has been growing interest in V2X research incorporating ISAC technology. There has been research on integrating sensors such as on-board units within vehicles and roadside units into ISAC-supported V2X networks to provide high-quality communication services [59]. The combination of RSMA’s ability to flexibly mitigate interference along with ISAC’s dual-functional capabilities makes RSMA-assisted V2X-ISAC a very interesting research area. In [79], a joint computation offloading and resource allocation strategy was investigated to make a reliable and efficient V2X network with mobile-edge computing (MEC) and ISAC technologies.

5. Conclusions

In this paper, we have explored the main technical challenges and potential research directions for RSMA-assisted ISAC systems. We have identified several prime technologies including NTN, RIS, mmWave and THz technologies, and V2X, and their roles in enhancing the performance and adaptability of RSMA-assisted ISAC systems. With these disruptive technologies, RSMA-assisted ISAC systems are envisioned to achieve ubiquitous global connectivity and precise target detection on the same platform with improved spectral and energy efficiency. We anticipate that this will bring a truly hyper-connected society and advanced systems capable of simultaneously performing communication and radar. While we find these technologies to be quite promising and relevant for RSMA-assisted ISAC networks, in the future, we will explore more technologies, and our perspectives on the significance and effectiveness of various technologies will naturally evolve. An important future direction is to consider AI and machine learning, which enable intelligent resource allocation and self-optimizing network behavior, as well as explore how the potential benefits of prime technologies enable various envisioned use cases and applications, including public safety, industrial automation, autonomous vehicles, smart cities, healthcare, and environmental monitoring, and many more.