Collision Resolution in Wireless Networks.

摘要

In a wireless uplink, collisions occur when two or more wireless users transmit signals at the same time over the same channel. Traditionally, when this happens, the received packets are discarded and retransmissions are required, which is a waste of power and bandwidth. The main contributions of this dissertation include a study of approaches for collision resolution, i.e., recovery of collided packets, the design of pulse-shape functions that facilitate collision resolution and also the analysis of packet delays in a cellular wireless network whose base station has collision resolution capability.;In the first part of this dissertation a new scheme, namely ALOHA with Collision Resolution (ALOHA-CR), is proposed, which is a cross-layer approach for high throughput wireless communications in a cellular uplink scenario. Transmissions occur in a time-slotted ALOHA-type fashion but with an important difference: simultaneous transmissions of two users can be successful. When two users transmit, the collision is resolved by oversampling the collision signal and exploiting independent information about the users that is contained in the signal polyphase components. The performance of ALOHA-CR is demonstrated on the Wireless Open Access Research Platform (WARP) testbed containing five software defined radio (SDR) nodes. The testbed results indicate that ALOHA-CR. leads to significant increase in throughput and reduction of service delays as compared to ALOHA. The second part of this dissertation focuses on optimal pulse-shape design for collision resolution. As mentioned above, the collided packets are separated by oversampling the collision signal. Because of oversampling, high correlations can occur between the columns of the virtual multiple-input multiple-output (MIMO) system matrix, which can be detrimental to user separation. A novel pulse-shape waveform design is proposed, which results in low correlation between the columns of the system matrix, while it exploits all available bandwidth as dictated by a spectral mask. In the third part, we study the delay properties of random scheduling (RS) in a cellular wireless network, under the assumption that the base station (BS) has multi-packet reception (MPR) capability. We minimize the expected delay of RS by determining the scheduling probabilities of nodes that will transmit simultaneously. For the perfect reception case, (i.e., when the success probability of transmissions is 1), a lower bound of the delay performance for an arbitrary scheduling policy is provided. The imperfect reception case is also studied and a convex optimization formulation is proposed, which can minimize the upper bound on the expected delay of RS. An approximation and a lower bound on the expected delay of RS are also developed under the assumption that the base station can support simultaneous transmission of two users.