HARMONIOUS MULTICAST RETRANSMISSION FOR LOW-POWER AND LOSSY NETWORKS FIRMWARE UPGRADING

Techniques are described herein for composing multicast retransmissions harmoniously by prioritizing transmission of new messages and regulating retransmission of old messages. Only the same multicast messages are forwarded simultaneously such that loss due to collision can be significantly reduced. By measuring the intervals of new messages and counting the duplicates, retransmissions are reasonably curbed with awareness of input rate and medium usage. This prevents the “domino effect” on a crowded channel when loss occurs. Moreover, self-silence mechanisms allow regular nodes to release channel resources for critical forwarders. Multicast retransmission is provided with guaranteed delivery rate, which is imperative for firmware upgrading in Low-Power and Lossy Networks (LLNs). DETAILED DESCRIPTION To perform large-scale firmware upgrading in Low-Power and Lossy Networks (LLNs), multicast is deemed a promising starting point to disseminate firmware blocks among mesh nodes. Minimizing the makespan of this upgrading procedure accelerates service recovery. To achieve this goal, many solutions have been proposed to enhance multicast delivery performance. Most of these solutions have so far been focused on source rate control, retransmission, or suppression schemes. Multicast Protocol for LLNs (MPL) uses proactive or on-demand multicast retransmission to combat packet loss. Meanwhile, trickle behavior is employed in MPL to suppress unnecessary retransmissions. However, multicast in LLNs suffers packet loss due to crowded broadcast channels. Though rate control can help alleviate a mild channel jam, it could also result in low 2 Chen et al.: HARMONIOUS MULTICAST RETRANSMISSION FOR LOW-POWER AND LOSSY NETWO Published by Technical Disclosure Commons, 2018 Copyright 2018 Cisco Systems, Inc. 2 5577 utilization and fail to take effect when multicast retransmission exists. Since the medium would again be packed with retransmitted packets, loss rate due to collision remains high, even if Clear Channel Assessment (CCA) is enabled. Figure 1 below illustrates a first example in which a new multicast and multicast retransmission occur concurrently. As shown, node A is broadcasting a new multicast packet while node C is retransmitting an old multicast packet. Node B would probably not detect either packet, and would need to issue a request for retransmission.