Marc E. Mosko

Custodian-based information sharing

Information sharing systems such as iCloud, Dropbox, Facebook, and Twitter are ubiquitous today, but all of them depend on massive server infrastructure and always-on Internet connectivity. We have designed and implemented a sharing system that does not require infrastructure yet supports robust, distributed, secure sharing by opportunistically using any and all connectivity, local or global, permanent or transient, to communicate. One key element of this system is a new information routing model that so far has proven to be as scalable and efficient as the best of the current Internet routing protocols, while operating in an environment more complex and dynamic than they can tolerate. The new routing model is made possible by new affordances offered by information-centric networking, in particular, the open source CCN [1] release. This article describes the new system and its routing model, and provides some performance measurements.

A new approach to on-demand loop-free routing in ad hoc networks

A new protocol is presented for on-demand loop-free routing in ad hoc networks. The new protocol, called labeled distance routing (LDR) protocol, uses a distance invariant to establish an ordering criterion and per-destination sequence numbers to reset the invariant resulting in loop-freedom at every instant. The distance invariant allows nodes to change their next hops or distances to destinations without creating routing-table loops. The destination sequence number, which only the destination may increment, permits nodes to reset the values of their distance invariants. The performance of LDR is compared against the performance of three other protocols that are representative of the state-of-the art, namely AODV, DSR and OLSR.

A self-correcting neighbor protocol for mobile ad-hoc wireless networks

Mobile wireless ad-hoc networks lack some basic abilities taken for granted in wired networks, such as the ability to know adjacent nodes. We present a neighbor discovery protocol, with particular application to broadcast flooding. The neighbor exchange protocol (NXP) has two main improvements over simple periodic broadcast schemes: (1) it only sends Hello packets when necessary to maintain topology and (2) uses sequence numbers in redistributed information to aid in convergence. In simulation, we compare NXP to a periodic protocol and simple flooding for all-node packet broadcasts and two dissemination techniques. We show that we maintain similar delivery rates while using fewer control packets in most configurations.

Distribution of route requests using dominating-set neighbor elimination in an on-demand routing protocol

The use of dominating-set neighbor elimination as an integral part of the distribution of route requests using the ad hoc on-demand distance vector (AODV) protocol as an example of on-demand routing protocols is investigated. We use detailed simulations to show that simply applying dominant pruning (DP) to the distribution of route requests in AODV results in pruning too many route requests in the presence of mobility and cross-traffic. Accordingly, we introduce several heuristics to compensate the effects of DP and show that the resulting AODV with dominating set heuristics (AODV-DS) has comparable or better delivery ratio, network load, and packet latency than the conventional AODV. AODV-DS exhibits over 70% savings on RREQ traffic than conventional AODV, and in some situations, AODV-DS may have a lower control overhead using Hello packets than conventional AODV without Helios.

An analysis of packet loss correlation in FEC-enhanced multicast trees

We study group loss probabilities of forward error correction (FEC) codes in shared loss multicast communication. We present a new analysis model with explicit state equations using recursive formulae. Our method looks at an FEC group as a whole, rather than analyze the number of transmissions of a particular packet. Our work applies to C(n,k) erasure codes, where any k out of n packets may decode the entire group. We find the cumulative distribution function that all leaf nodes in a shared loss tree successfully decode a C(n,k) FEC group, the probability mass function (p.m.f.) for the number of leaf nodes that successfully decode a transmission group, the expected number of packets received on successful decode and the expected number of missing packets on decode failure for a particular leaf node of the multicast tree, the p.m.f. that all leaf nodes hold the same packets in common, and the expected height of packet loss. Most of our findings generalize to arbitrary trees with non-uniform link loss. Our results also apply to non-FEC trees. We illustrate applications of our work with examples.

Performance of group communication over ad-hoc networks

We study the performance of reliable and unreliable all-node broadcast over ad-hoc networks that use contention-based channel access. To obtain analytical results while preserving hidden-terminal and node clustering characteristics of ad-hoc networks, we introduce a novel differential-equation fluid model for information flow through a network of cluster trees, where a spanning tree joins groups of fully connected nodes. Through numerical analysis and simulations in GloMoSim, we show throughput, goodput, and loss rates for reliable and unreliable networks. For reliable broadcast, we also find NAK rates, NAK loss rates, and retransmission rates. We show that using end-to-end sequence numbers, which are common in reliable multicast, for NAK generation in ad-hoc networks creates substantial unnecessary traffic.

Using utility and microutility for information dissemination in Vehicle Ad Hoc Networks

We describe an approach to propagating streams of information in Vehicle Ad-Hoc Networks (VANETs) based on sources of information anticipating where their information will be useful. In this paper we describe how sources can model the potential usefulness of their information using utility functions. These utility functions are converted to more compact ldquomicroutilitiesrdquo that travel with the individual data packets. The microutilities allow the information forwarding protocols to operate distributedly and independently on individual data packets, while achieving good overall coordination and delivery for entire data streams. We describe the algorithms used to convert utility functions to microutilities. Our algorithms insure that both proactive planning and reactive dropping of information in-transit are done consistent with the needs of the different applications. In this way data streams from both high priority (safety) and lower priority (traffic and commercial) applications can be propagated in the same network. We show experimental results that demonstrate the advantage of such a utility.microutility approach in serving the needs of diverse intelligent transportation system applications.

A new approach to on-demand loop-free routing in networks using sequence numbers

A new protocol is presented for on-demand, loop-free routing in ad hoc networks. The new protocol, called the labeled distance routing (LDR) protocol, uses a distance invariant to establish an ordering criterion and per-destination sequence numbers to reset the invariant resulting in loop-freedom at every instant. The distance invariant allows nodes to change their next hops or distances to destinations without creating routing-table loops. The destination sequence number, which only the destination may increment, permits nodes to reset the values of their distance invariants. The performance of LDR is compared against the performance of three other protocols that are representative of the state-of-the-art, namely AODV, DSR and OLSR; LDRs performance is shown to be far better than the other three protocols.

Ad Hoc Routing with Distributed Ordered Sequences

Ad hoc routing with distributed ordered sequences Marc Mosko ∗ Palo Alto Research Center 3333 Coyote Hill Road Palo Alto, CA 94304 Email: mmosko@parc.com Abstract— We propose a new hop-by-hop routing protocol for ad hoc wireless networks that uses a novel sequence number scheme to ensure loop-freedom at all times. We use a single large per-destination label space to order nodes in a topological sort (directed acyclic graph). Nodes manipulate the label set in- network without needing destination-controlled resets, so path repair is localized. The label size is large enough that it should never be exhausted in the lifetime of any given network. Route request flooding is performed through a new method that exploits the inherent partial order of the network, so nodes can share RREQ floods. Whereas most previous route request pruning techniques create a request tree, the new technique creates a directed acyclic request graph. Simulation results compared to AODV, DSR and OLSR show that the new protocol has in most cases equivalent or better packet delivery ratio and latency, but with a fraction of the network load. I. I NTRODUCTION Wireless ad hoc computer networks are communications networks in which each node may be mobile and has at least one radio interface. There is no central infrastructure, such as cell towers, base stations, for access points. Exam- ples of these networks include tactical military applications, commercial vehicle-to-vehicle systems such as DSRC [1], or emergency rescue impromptu networks. The Internet Engi- neering Task Force (IETF) studies such networks under the mobile ad hoc networks (MANET) working group. Three MANET routing protocols have request for comments (RFC) status and two have internet draft (ID) status. The three MANET RFC protocols are the Adhoc On-demand Distance Vector (AODV) routing protocol [2], the Optimized Link State Routing (OLSR) protocol [3], and the Topology Dissemination Based on Reverse-Path Forwarding (TBRPF) [4]. The two Internet Drafts are the Dynamic Source Routing Protocol (DSR) [5], and the Dynamic MANET On-demand (DYMO) Routing [6]. TBRPF and OLSR are examples of link-state protocols, where nodes exchange topology information and execute a shortest path algorithm (e.g. Dijkstra’s) using the topology information they maintain to find routing paths. Unfortunately, neither of these protocol are loop-free, which means that the routing tables at nodes may point in a directed cycle at times. AODV, DSR, DYMO operate as on-demand protocols, which means that they do not maintain routes for all destinations, only those for which there is traffic. Nodes discover paths in on-demand routing protocols through route request (RREQ) floods in the network and unicast route reply J.J. Garcia-Luna-Aceves ∗† Computer Engineering Department University of California at Santa Cruz Santa Cruz, CA 95064 Email: jj@soe.ucsc.edu (RREP) advertisements. The IETF on-demand protocols attempt to maintain loop- free operation through different techniques. DSR is a source- routing protocol, so each source node must maintain complete path information to each in-use destination. If there are path changes, then the protocol must either drop the traffic or use a recovery technique, which has been shown to be prone to looping. AODV uses distance labels (hop count) to order nodes along shortest paths. If a node needs to repair a path, it increments a destination sequence number and broadcasts a RREQ. By incrementing the sequence number, it prevents any predecessors (nodes that use the current node as a suc- cessor) from replying and maintains loop-freedom. DYMO also uses distance labels and sequence numbers to maintain loop-freedom. All RREQ broadcasts must be answered by the destination node, and the destination will increase a route sequence number if the requested sequence number is larger than the stored number, or the reply path length is longer than the requested path length. Jaffe and Moss [7] made a key observation in the study of loop-free distance vector routing protocols. They noted that a node may independently add a new successor to a destination if the new distance does not exceed the current distance. If the distance increases, then the node must coordinate with other nodes through some mechanism. The coordination must ensure both that the new successor path is loop-free and that the new distance at the node is not out-of-order with respect to any predecessors. The conditions in DUAL [8] generalize the work of Jaffe and Moss and provide a reliable mechanism to reset the ordering information at predecessors through a diffusing computation [9]. Reliable diffusing computations, however, are impractical in a wireless ad hoc network due to their overhead and latency problems timing out non-existent links. Loop-free ad hoc routing protocols address the reset condi- tion several ways. One approach is to use source routing, such as in DSR and variations on it. Another approach consists of relying on reliable internodal coordination based on the values of distances to destinations reported by nodes (e.g., DUAL [8], LPA [10], ROAM [11]). Yet another approach, used in proto- cols similar to AODV, consists of having nodes use a distance label (D) and a sequence number (SN) to reset distance values. In this case, The ordered pair (SN, D) constitutes a lexico- graphic order. AODV manipulates the pair such that when there is a link break a node requests the next higher SN. This 1-4244-0222-0/06/

Secure Fragmentation for Content Centric Networking

20.00 (c)2006 IEEE This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the Proceedings IEEE Infocom.