Xiangping Qin

Opportunistic splitting algorithms for wireless networks

In this paper, we develop medium access control protocols to enable users in a wireless network to opportunistically transmit when they have favorable channel conditions, without requiring a centralized scheduler. We consider approaches that use splitting algorithms to resolve collisions over a sequence of minislots, and determine the user with the best channel. First, we present a basic algorithm for a system with i.i.d. block fading and a fixed number of backlogged users. We give an analysis of the throughput of this system and show that the average number of minislots required to find the user with the best channel is less than 2.5 independent of the number of users or the fading distribution. We then extend this algorithm to a channel with memory and also develop a reservation based scheme that offers improved performance as the channel memory increases. Finally we consider a model with random arrivals and propose a modified algorithm for this case. Simulation results are given to illustrate the performance in each of these settings

Opportunistic Splitting Algorithms for Wireless Networks with Fairness Constraints

In wireless networks, it is well established that the throughput can be increased by opportunistically scheduling transmissions to users that have good channel conditions. Several “opportunistic” medium access control protocols have been developed, which enable distributed users to opportunistically transmit without requiring a centralized scheduler. In this paper, we consider opportunistic splitting algorithms, where a sequence of mini-slots is used to determine the appropriate user to schedule at each time. In prior work, this type of algorithm has been developed for homogeneous systems in which all users have independent and identically distributed (i.i.d.) channel statistics. Here, we specify new splitting algorithms for a heterogeneous environment that may also include fairness constraints. The performance of the splitting algorithms are characterized via analysis and simulations. In particular, we show that in certain cases, a heterogeneous algorithm will perform at least as well as the homogeneous algorithm in a system with the same total number of users.

A distributed splitting algorithm for exploiting multiuser diversity

Multiuser diversity [1] refers to the inherent diversity present across the user population in a wireless network. For example, in a fading multiaccess channel, this diversity can be exploited by allowing only the user with the best channel conditions to transmit at any time. However, this requires a centralized scheduler with knowledge of each user’s channel gain. Such an approach may not scale well for large networks, and the delays involved in gathering the required knowledge may also limit the performance. In this paper, we focus on decentralized approaches for exploiting multiuser diversity, where each user only has knowledge of its own fading level, but no knowledge of the fading levels of the other users in the cell. This is similar to decentralized power control problems, as in [2]. In [4], we have shown that multi-user diversity can still be exploited in a distributed setting by using a simple variation of a slotted ALOHA protocol, where each user randomly transmits based on their local channel knowledge. By appropriate choice of transmission probability, it can be shown that, asymptotically, the only penalty incurred from distributed channel knowledge is the contention inherent in the ALOHA protocol. Also, with a moderate number of backlogged users this approach achieves a higher throughput than a deterministic TDMA protocol. In this paper, we consider a type of splitting algorithm [3] to reduce the contention when the timescale over which the channel varies is larger than the roundtrip time between each transmitter and the receiver. Unlike traditional splitting algorithms, the goal is not to simply allow all backlogged users to transmit, but to enable the user with the best channel conditions to transmit. We consider the time-slotted block-fading model where the channel gain is fixed during each slot and changes independently between slots. At the beginning of each time-slot, several mini-slots of length β are used to execute the splitting algorithm. We assume that each time-slot contains K minislots. The splitting algorithm will determine two thresholds, Hl and Hh for each mini-slot, such that only users whose channel gains, h satisfy Hl < h < Hh are allowed to transmit. After a mini-slot, each user receives (0, 1, e) feedback, indicating if the transmissions in a mini-slot resulted in an idle, success or collision. We denote the received feedback by m. If m = 1, only the user with best channel gain transmitted in the mini-slot; the user will then continue to transmit through the remainder of the time slot. If m = 0 or m = e, the users will adjust their thresholds and repeat the algorithm until either a success occurs or the time-slot ends. The exact manner in which this is done is given by the following pseudo-code. Here Hll is largest value of Hl used in prior mini-slots such that there are some users above Hll.

Distributed Resource Allocation and Scheduling in OFDMA Wireless Networks

In this paper we develop distributed resource allocation and scheduling algorithms for the uplink of an orthogonal frequency division multiple access (OFDMA) wireless network. We consider a time-slotted model, where in each time-slot the users are assigned to subchannels consisting of groups of OFDM tones. Each user can also allocate its transmission power among the subchannels it is assigned. We consider distributed algorithms for accomplishing this, where each users actions depend only on knowledge of their own channel gains. Assuming a collision model for each subchannel, we characterize an optimal policy which maximizes the system throughput and also give a simpler sub-optimal policy. We study the scaling behavior of these policies in several asymptotic regimes for a broad class of fading distributions.

SINR-based channel assignment for dense wireless LANs

The biggest challenge in channel assignment for dense, multi-cell/AP wireless LANs is to arrange cochannel cells so as to maximize the aggregate network throughput. Most previous work models this problem as a vertex coloring problem. In this paper we model it as a non-linear optimization problem to maximize overall network throughput. We prove that the new optimization problem is NP-hard and vertex-coloring is a simplified case. We then propose a polynomial time heuristic algorithm called MIF (most-interfered-first) for channel assignment. The performance for a line topology is analyzed. Simulations for random topologies show that MIF consistently produces better network throughput than vertex-coloring based heuristic algorithms with less computation cost.