While routing in wireless networks has been studied extensively, existing protocols are typically designed for a specific set of network conditions and so do not easily accommodate changes in those conditions. For instance, protocols that assume network connectivity cannot be easily applied to disconnected networks. In this paper, we develop a distributed routing strategy based on deep reinforcement learning that generalizes to diverse traffic patterns, congestion levels, network connectivity, and link dynamics. We make the following key innovations in our design: (i) the use of relational features as inputs to the deep neural network approximating the decision space, which enables our algorithm to generalize to diverse network conditions, (ii) the use of packet-centric decisions to transform the routing problem into an episodic task by viewing packets, rather than wireless devices, as reinforcement learning agents, which provides a natural way to propagate and model rewards accurately during learning, and (iii) the use of extended-time actions to model the time spent by a packet waiting in a queue, which reduces the amount of training data needed and allows the learning algorithm to converge more quickly. We evaluate our routing algorithm using a packet-level simulator and show that the policy our algorithm learns during training is able to generalize to larger and more congested networks, different topologies, and diverse link dynamics. Our algorithm outperforms shortest path and backpressure routing with respect to packets delivered and delay per packet.

Detecting node failures in mobile wireless networks is very challenging because the network topology can be highly dynamic, the network may not be always connected, and the resources are limited. In this paper, we take a probabilistic approach and propose two node failure detection schemes that systematically combine localized monitoring, location estimation and node collaboration. Extensive simulation results in both connected and disconnected networks demonstrate that our schemes achieve high failure detection rates (close to an upper bound) and low false positive rates, and incur low communication overhead. Compared to approaches that use centralized monitoring, our approach has up to 80 percent lower communication overhead, and only slightly lower detection rates and slightly higher false positive rates. In addition, our approach has the advantage that it is applicable to both connected and disconnected networks while centralized monitoring is only applicable to connected networks. Compared to other approaches that use localized monitoring, our approach has similar failure detection rates, up to 57 percent lower communication overhead and much lower false positive rates (e.g., 0.01 versus 0.27 in some settings).

Most existing studies on Delay/Disruption Tolerant Networks (DTNs) consider pair-wise node encountering that assumes nodes only meet in pairs. In many mobile wireless networks, a group of nodes, instead of only a pair of nodes, may meet each other. In this paper, we study how to effectively transmit a set of packets from a source to a destination in such group meeting scenarios. The optimization goal is to minimize the delay for the packets to reach the destination while limiting the energy consumption. We first assume that node encountering is known beforehand, and develop an algorithm to obtain the minimum delay. We then develop two practical network coding based schemes. Both schemes use a token technique to limit the total number of transmissions, and only incur signaling at the beginning of a group meeting. One scheme requires nodes in a group to exchange their encoding matrices with each other, while the other only requires exchanging rank information. Simulation results demonstrate that both schemes achieve delays close to the minimum delay for moderate number of tokens. They present different tradeoffs in the number of transmissions and the signaling overhead.

In this paper, we study data gathering in wireless sensor networks where intermediary nodes perform data aggregation with total fusion. We focus on gathering tree construction (from a general network topology) and transmission scheduling in order to minimize gathering delay. We first propose an algorithm that calculates an optimal transmission schedule and the minimal delay, for any given gathering tree. Then, we prove a lower bound, in terms of the number of nodes in the network, on the optimal gathering delay for any graph. After analyzing the gathering trees formed by several popular tree constructing algorithms, we propose an algorithm that constructs the optimal gathering tree for a complete graph. We then conduct extensive simulations to show that the proposed algorithm is also a promising approximation algorithm for arbitrary graphs. We also propose an approximation algorithm that constructs a gathering tree that achieves a maximum-degree-optimal solution.

In this paper, we investigate the problem of determining the routing that minimizes the maximum/average delivery time or the maximum/average delivery delay for a set of packets in a deterministic Delay Tolerant Network, i.e. in a network for which all the nodes' transmission opportunities are known in advance. While the general problem with multiple sources and multiple destinations is NP-hard, we present a polynomial time algorithm that can efficiently compute the optimal routing in the case of a single destination or of a single packet that needs to be routed to multiple destinations.

In this paper, we investigate the benefits of applying a form of network coding known as random linear coding (RLC) to unicast applications in disruption-tolerant networks (DTNs). Under RLC, nodes store and forward random linear combinations of packets as they encounter each other. For the case of a single group of packets originating from the same source and destined for the same destination, we prove a lower bound on the probability that the RLC scheme achieves the minimum time to deliver the group of packets. Although RLC significantly reduces group delivery delays, it fares worse in terms of average packet delivery delay and network transmissions. When replication control is employed, RLC schemes reduce group delivery delays without increasing the number of transmissions. In general, the benefits achieved by RLC are more significant under stringent resource (bandwidth and buffer) constraints, limited signaling, highly dynamic networks, and when applied to packets in the same flow. For more practical settings with multiple continuous flows in the network, we show the importance of deploying RLC schemes with a carefully tuned replication control in order to achieve reduction in average delay, which is observed to be as large as 20% when buffer space is constrained.

We study traces taken from UMass DieselNet, a Disruption-Tolerant Network consisting of WiFi nodes attached to buses. As buses travel their routes, they encounter other buses and in some cases are able to establish pair-wise connections and transfer data between them. We analyze the bus-to-bus contact traces to characterize the contact process between buses and its impact on DTN routing performance. We find that the all-bus-pairs aggregated inter-contact times show no discernible pattern. However, the inter-contact times aggregated at a route level exhibit periodic behavior.Based on analysis of the deterministic inter-meeting times for bus pairs running on route pairs, and consideration of the variability in bus movement and the random failures to establish connections, we construct generative route-level models that capture the above behavior. Through trace-driven simulations of epidemic routing, we find that the epidemic performance predicted by traces generated with this finer-grained route-level model is much closer to the actual performance that would be realized in the operational system than traces generated using the coarse-grained all-bus-pairs aggregated model. This suggests the importance in choosing the rightlevel of model granularity when modelingmobility-related measures such as inter-contact times in DTNs.

We present the design, implementation, and performance evaluation of AMPS --- a flexible, scalable proxy testbed that supports a wide and extensible set of next-generation proxy streaming services. AMPS employs a modular architecture and is built on top of a commodity Linux system. We study the performance of AMPS proxy using a server-proxy-client configuration in a switched-Gigabit LAN environment. We identify the CPU to be the system bottleneck. Through profiling study, we further identify the kernel network protocol processing and the Network Reception Module inside the proxy to be the most CPU-intensive components. We also quantify the maximum achievable throughput for two of the principal components of the proxy - the control plane and data plane, and characterize the end-to-end performance along the server-to-proxy-to-client path. We discuss lessons learned and the various optimizations made in the course of our study to improve system performance.