Orchestrating Spatial Reuse in Wireless Ad Hoc Networks Using Synchronous Collision Resolution (Scr)

Abstract

We propose a novel medium access control protocol for ad hoc wireless networks data to send can contend simultaneously for the channel. Nodes contend for access using a synchronous signaling mechanism that achieves two objectives: it arbitrates contentions locally and it selects a subset of nodes across the network that attempt to transmit simultaneously. The subset of nodes that survive the signaling mechanism can be viewed as an orchestrated set of transmissions that are spatially reusing the channel shared by the nodes. Thus the 'quality' of the subset of nodes selected by the signaling mechanism is a key factor in determining the spatial capacity of the system. In this paper, we propose a general model for such synchronous signaling mechanisms and recommend a preferred design. We then focus via both analysis and simulation on the spatial and capacity characteristics of these access control mechanisms. Our work is unique in that it specifically focuses on the spatial capacity aspects of a MAC protocol, as would be critical for ad hoc networking, and shows SCR is a promising solution. Specifically, it does not suffer from congestion collapse as the density of contending nodes grows, it does not suffer from hidden or exposed node effects, it achieves high capacities with a spatial usage exceeding 1 (i.e. more than one packet exchange in the area covered by a transmission), and it facilitates the integration of new physical layer capacity increasing technologies.

Modelling Throughput and Starvation in 802.11 Wireless Networks with Multiple Flows

“…The three-flows problem and variations have been considered by other researchers under several names, e.g., the"exposed terminal problem"( [4], [5]). The issue of asynchronization as a cause of unfairness under 802.11 has been noted specifically in [6] and [7].…”

“…A set of nodes that can transmit simultaneously forms an "independent set" ( [9], [12], [10], [15]) in the conflict graph; an independent set which can't have any more nodes added to it is maximal. In Figure 3a, the maximal independent sets are (1,3,5), (1,4), (2,5), and (2,4). In Figure 3b, the maximal independent sets are (1,4), (1,5), (2,5) and (3).…”

“…For instance, in the n = 5, k = 2 topology, state (1,5) is equally likely to transition into states (1,4) and (2,5). In the n = 6, k = 2 topology, state (1,6) is equally likely to transition into states (1,5), (1,4), (2,6), or (3,6).…”

“…In the case where k + 1 does not divide n−1, there are more than one largest state. For example, if n = 8 and k = 2, the size L = 3 sets are (1, 4, 7), (2,5,8), (1,4,8), and (1, 5, 8). Since we made the argument that all size-L states occur an equal fraction of the time, the long-run fraction of time that a node i gets the channel is given by the fraction of such states (size L maximal independent sets) that i belongs to:…”

A Novel MAC Protocol for Wireless Ad Hoc Networks with Power Control

“…CRS consists of consecutive signaling phases and assertion signals, and the signaling phase is designed by choosing the number of signaling slots and signal selection probabilities [8] [9]. A node survives CRS by surviving all signaling phases.…”