Current Research Interests

A Coherent Theory of Random Access

As one of the two fundamental types of multiple access, random access has been widely adopted in various communication networks, and expected to play an increasingly central role owing to the rising popularity of Machine-to-Machine (M2M) communications. Studies on random access date back to 1970s. After decades of extensive research, numerous random-access schemes have been developed and successfully adopted in various wireless communication networks including cellular networks and IEEE 802.11 (Wi-Fi) networks.

In stark contrast to the wide applications, the theory of random access is, nevertheless, much less developed. Analytical models are often tailored for specific performance metrics or access protocols, where differences in modeling assumptions and definitions have led to inconsistent findings. Despite continuous attention for almost half a century, many fundamental issues remain unresolved.

Essentially, a random-access network can be regarded as a multi-queue-single-server system, where the service processes of nodes' queues are coupled not only with each other, but also with their arrival processes. The challenge of establishing a coherent theory for random-access networks originates from the lack of characterization of service rates of coupled queues. In our recent work [Dai'22], we addressed the three key issues for stability of random-access networks, that is, 1) how to characterize the coupled service rates, 2) how to determine the stability region of input rates, only within which the netweork can be stabilized, and 3) for given input rates within the stability region, how to tune the transmission probabilities of nodes to stabilize the network.

[Dai'22] is indeed an extension from our previous work on sensing-free [Dai'12] and sensing-based random access [Dai'13], where a unified analytical framework was proposed for the symmetric scenario, i.e., all the nodes have identical traffic statistics and access parameters.  It has been elaborated in a series of papers including [Li-Dai'16], [Sun-Dai'16], [Sun-Dai'17], [Li-Dai'18], [Gao-Dai'19], [Sun-Dai'19] and [Gao-Fang-Song-Dai'22], and successfully applied to Wi-Fi networks in [Dai-Sun'13, Gao-Sun-Dai'13, Gao-Dai'13, Gao-Sun-Dai'14, Sun-Dai'15, Gao-Dai-Hei'17, Gao-Sun-Dai'19] and LTE networks in licensed bands [Zhan-Dai'18], [Zhan-Dai'19] and [Zhan-Dai'19-WCL] and unlicensed bands [Sun-Dai'20].

With the proposed unified analytical framework, a series of fundamental performance limits of random access can be characterized, along with the optimal setting of access parameters for approaching the limiting performance.  It not only provides direct guidance to the performance optimization of existing access protocols, but also sheds important light on the access design of next-generation communication networks.

Fundamental Theory of Random Access

* A Unified Analytical Framework

* Maximum Sum Rate

* Packet-Based versus Connection-Based

* To Sense or Not to Sense

Performance Optimization of IEEE 802.11 DCF Networks

* A Unified Analysis: Stability, Throughput and Delay

* IEEE 802.11e EDCA: Modeling, Differentiation and Optimization

* Backoff Design: Fundamental Tradeoff and Design Criterion

* Multi-BSS with Universal Frequency Reuse

* Multi-Standard IEEE 802.11 Networks

Massive Random Access of M2M Communications in LTE Networks

* Throughput Optimization

* Delay Optimization

Optimal Decomposition for Large-Scale Infrastructure-Based Wireless Networks

The fundamental idea of network decomposition is to break a large-scale network into smaller parts such that the subnetworks can operate in parallel, each with a much lower dimensionality. For large-scale wireless networks, the cellular structure is based on the idea of network decomposition, where the network is decomposed into multiple subnetworks, i.e., cells, according to the coverage of each base-station (BS). Such a decomposition scheme, nevertheless, leads to strong interference among subnetworks, which becomes increasingly significant as the density of BSs grows. For the next-generation cellular network where a massive amount of BSs need to be deployed to meet the ever-increasing demand of high data rate, it is of paramount importance to develop efficient network decomposition schemes to replace the current cellular structure. How to build such a decomposition framework, unfortunately, has remained largely unknown.

In our recent work [Dai-Bai'17], a network decomposition theory is established for large-scale wireless networks from a graph-theoretic point of view. Specifically, we start from a novel bipartite graph representation of an infrastructure-based wireless network, and show that in general the optimal network decomposition can be formulated as a graph partitioning problem. For demonstration, we focus on maximizing the number of subgraphs for a given cut ratio constraint, and propose a Binary Search based Spectral Relaxation (BSSR) algorithm to solve it in two loops. The performance of the proposed BSSR algorithm is further examined and compared to the current cellular structure and BS clustering in various scenarios. Significant gains are shown to be achieved by the proposed BSSR algorithm, which corroborates that the optimal network decomposition of next-generation cellular networks should be performed based on a bipartite graph where the geographical information of BSs and users are both included.

Modeling and Performance Analysis of Large-Scale Distributed Antenna Systems (DASs)

The distributed antenna system (DAS) has become a promising candidate for next-generation (5G) mobile communication systems. In contrast to the conventional cellular structure where antennas are co-located at the tower-mounted base station (BS) in each cell, in a DAS, many low-power remote antenna ports are geographically distributed over a large area and connected to a central processor by fiber. The appealing features of distributed antennas have attracted considerable attention from both industry and academia, and been applied to the cutting-edge technologies such as small cells and the Cloud Radio Access Network (C-RAN).

In the next-generation mobile communication systems, a large amount of BS antennas are expected to be deployed to meet the ever increasing demand of high data rate. Significant efforts have been made on the performance analysis of cellular systems with large antenna arrays at BSs (popularly known as “massive MIMO”). If the BS antennas are distributed, on the other hand, how the capacity scales with the number of BS antennas is less clear. In our recent work, a comprehensive comparison on the capacity scaling laws of MIMO cellular systems with co-located and distributed BS antennas is presented for both uplink [Dai'11, Dai'14] and downlink [Liu-Dai'14, Wang-Dai'15]. The asymptotic analysis shows that the scaling order is crucially dependent on the BS antenna layout. If the number of BS antennas and the number of users both go to infinity but their ratio is fixed, for instance, the uplink sum capacity [Dai'14] and the downlink sum rate with orthogonoal precoding schemes [Liu-Dai'14, Wang-Dai'15] converge to a constant with BS antennas co-located at the center of each cell. In contrast, with BS antennas uniformly distributed in each cell, the sum capacity/rate increases with the number of BS antennas unboundedly.

The analysis also shows that despite better capacity/rate performance, the cell-edge problem could be exacerbated if distributed BS antennas are used in cellular systems. As pointed out in [Dai'14], the cell-edge problem has its roots in the cellular structure where cells are formed based on the coverage of each BS. Such a BS-centric structure, nevertheless, cannot be justified when both users and BS antennas are scattered around. Instead, the signal processing may be performed based on the unit of “virtual cells”. It is shown in [Dai'14] that a uniform inter-cell interference density can be achieved in a DAS if each user chooses a few surrounding BS antennas to form its virtual cell. By doing so, each BS antenna serves a declining number of users as the density of BS antennas increases, indicating good scalability that is much appreciated for a large-scale network.

For virtual-cell based DASs, the virtual cell size, i.e., how many BS antennas should be included into each user's virtual cell, is a key system parameter. To study the effect of virtual cell size, [Wang-Dai'16] considered a large-scale downlink DAS with two representative linear precoding schemes: maximum ratio transmission (MRT) and zero-forcing beamforming (ZFBF). The analysis shows that if MRT is adopted in each user's virtual cell, a small virtual cell size should be chosen so as to avoid sharing BS antennas for different users which would otherwise cause strong interference. On the other hand, if users are grouped with joint ZFBF transmission from their virtual cells to eliminate the intra-group interference, the average user rate could be significantly improved by increasing the virtual cell size. A novel virtual-cell based user grouping algorithm is proposed, with which the rate difference among users is greatly reduced compared to the conventional BS-centric clustering.

Performance Comparison of Cellular Systems with Co-located BS Antennas and Distributed BS Antennas

* Uplink Sum Capacity

* Downlink Average User Rate with Linear Precoding

Performance Analysis and System Design for Virtual-cell based DASs