Scalability of Broadcast Performance in Wireless Network-on-Chip

Abstract

Abstract-Networks-on-Chip (NoCs) are currently the paradigm of choice to interconnect the cores of a chip multiprocessor. For hundreds or thousands of cores, though, conventional NoCs may not suffice to fulfill the increasing on-chip communication requirements given that the performance of such networks actually drops as the number of cores grows, especially in the presence of multicast and broadcast traffic. This not only limits the scalability of current multiprocessor architectures, but also sets a performance wall that prevents the development of architectures that generate moderate-to-high levels of multicast. In this paper, a Wireless Network-on-Chip (WNoC) where all cores share a single broadband channel is presented. Such design is conceived to provide low latency and ordered delivery for multicast/broadcast traffic, in an attempt to complement a wireline NoC that will transport the rest of communication flows. To assess the feasibility of this approach, the network performance of WNoC is analyzed as a function of the system size and the channel capacity, and then compared to that of wireline NoCs with embedded multicast support. Based on this evaluation, preliminary results on the potential performance of the proposed hybrid scheme are provided, together with guidelines for the design of MAC protocols for WNoC.

OrthoNoC: A Broadcast-Oriented Dual-Plane Wireless Network-on-Chip Architecture

“…Related works generally resort to contention-free mechanisms via multiplexing or variants of the popular token-passing protocol [7], [8], [36]. These methods do not scale well with the number of participating nodes due to the high cost of introducing new channels or the increase of the token round trip time [37]. Better scalability is achieve with the protocol used in [29], where nodes request access by broadcasting short orthogonal request packets.…”

“…For this, we adopt a family of protocols that let nodes to contend for the channel and resolve collisions in a distributed manner. The reason is that such contention-based protocols are generally more scalable than the contention-free alternatives and naturally adapt to hotspot traffic and other variations [37], as well as to changes in the number of available wireless interfaces. BRS-MAC [38], the protocol employed in ORTHONOC, maintains these advantages and minimizes the penalty of collisions via three techniques: preamble transmission, collision detection, and scalable acknowledging.…”

“…Evaluation metrics: On the one hand, we measure latency as the time passed between the generation of a message and its the complete reception at all the intended destinations. For simplicity, we report the average communication latency for low loads, as this models the performance of the network for at least 40% of the maximum admitted load with high accuracy [17], [37]. On the other hand, the throughput accounts for the aggregate of both the unicast and broadcast flows and is measured from the transmitter perspective, i.e., messages with multiple receivers are only counted once.…”

“…These proposals are not oriented to broadcast, which has been shown to be a strong advantage of WNoC in manycores [37]. In spite of this, a few works have discussed evaluated architectures with multicast capabilities.…”

Medium Access Control in Wireless Network-on-Chip: A Context Analysis

“…The use of coordinated access protocols such as token passing, either alone [3] or in combination with channelization techniques [4], alleviates these flexibility and scalability concerns. Still, the overhead associated to the token circulation may hinder the use of such technique in manycore environments [7]. On the other end, random access techniques provide flexible operation and low latency as nodes can attempt to gain access at any time instant.…”