Yuxi Li

Optimal Traffic-Oblivious Energy-Aware Routing for Multihop Wireless Networks

Energy efficiency is an important issue in multihop wireless networks with energy concerns. Usually it is achieved with accurate knowledge of the traffic pattern and/or the current network information such as the remaining energy level. We investigate the problem of designing a routing scheme to minimize the maximum energy utilization of a multihop wireless network with weak assumption of the traffic pattern and without ongoing collection of network information. We develop polynomial size LP models to design such a routing scheme. We discuss generalizations of the LP models to various radio transmission models. In an interference-limited scenario, we show how to guarantee schedulability of the oblivious routing. We present an extension to consider lossy links. We also discuss implementation issues. The LP models achieve performance close to what an oracle can achieve in the performance study. The results for multihop wireless networks with a single sink are especially good. We make a first stride in designing a traffic-oblivious energy-aware routing framework in multihop wireless networks.

Impact of lossy links on performance of multihop wireless networks

Multihop wireless networks have unique features such as lossy links and interference. Both interference and lossy links affect the maximum achievable throughput of a network. Some wireless networks have energy constraints. Lossy links also affect energy efficiency due to retransmissions and broadcasting. We investigate the impact of lossy links on maximum achievable throughput and minimax energy utilization. These can be modeled as linear programming optimization problems. We give optimal solutions for both flow-based and destination-based routing. Experiments show that lossy links do have significant impact on the maximum achievable throughput. There are cases where a network can only achieve half of the throughput of the corresponding lossless network. The results show less significant impact of loss on energy efficiency. In some cases, the loss may be advantageous for energy efficiency, since the energy consumption may be reduced due to the loss of broadcasting messages. Experiments also show the significant impact of overhearing on energy efficiency.

A simple method for balancing network utilization and quality of routing

Applegate and Cohen (2003) computational complexity design demand oblivious routing schemes that achieve low oblivious ratio with no or approximate knowledge of traffic demands. We investigate the quality of oblivious routing with respect to path dispersion, which is concerned with the number of paths; and path variation, which is concerned with how far the paths are from the shortest paths and the variation of path lengths. The results show the dispersion and the variation are high in general. We propose a penalty method to improve the quality of oblivious routing. The penalty method strikes a good balance between the conflicting objectives of minimizing the oblivious ratio and optimizing the quality of oblivious routing. Moreover, we apply the simple penalty method to the problem of minimizing the maximum link utilization given a traffic matrix. With the penalty method, we can achieve almost the same maximum link utilization, and improve the quality of routing to almost perfect, i.e. one or two paths that are very close to the shortest paths between each pair of nodes.

Traffic-oblivious Energy-aware Routing for Multihop Wireless Networks

Energy efficiency is an important issue in multihop wireless networks with energy concerns. Usually it is achieved with accurate knowledge of the traffic pattern and/or the current network information such as load and remaining energy level (in an energy constrained case). However, accurate estimation of the traffic pattern is hard to obtain. Transmission interference and lossy links exacerbate the problem. We have investigated the problem of designing a routing scheme to minimize the maximum energy utilization of a multihop wireless network with no or weak assumption of the traffic pattern and without ongoing collection of network information. We have developed polynomial size LP models to design such a traffic-oblivious energy-aware routing scheme, in scenarios where 1) there is no interference and links are lossless, and 2) interference is present and links are lossy. The experimental results show that the oblivious routing achieves the performance close to what an oracle can achieve. The results for multihop wireless networks with a single sink are especially good.

Stable and robust multipath oblivious routing for traffic engineering

Intra-domain traffic engineering is essential for the operation of an Internet Service Provider. Demand-oblivious routing [2] promises excellent performance guarantee with changing and uncertain traffic demands. However, it is difficult to implement it. We investigate an efficient and deployable implementation of oblivious routing. We study its performance by both numerical experiments and simulation. The performance study shows that the multipath implementation achieves a close approximation to oblivious routing [2], especially when approximate knowledge of traffic is available. The study also shows its robustness under varying traffic demands, link failures and an adversary attack. Its performance is excellent even with a 100% error in traffic estimation.

Multipath Oblivious Routing for Traffic Engineering - Stable and Robust Routing in Changing and Uncertain Environments

Intra-domain traffic engineering is essential for the operation of an Internet Service Provider (ISP). Demandoblivious routing [5] promises excellent performance guarantee with changing and uncertain traffic demands. However, it is non-trivial to implement it. We investigate an efficient and deployable implementation of oblivious routing. We study its performance by both numerical experiments and simulation. The performance study shows that the multipath implementation achieves a close approximation to oblivious routing [5], especially when approximate knowledge of traffic is available. The study shows its robustness under varying traffic demands, link failures and an adversary attack. Its performance is excellent even with a 100% error in traffic estimation. We open the door for a deployable demand-oblivious routing, which can provide robust network services with good quality to network users.

Active error recovery for reliable multicast

An error recovery scheme is essential for large-scale reliable multicast. We design, implement, and evaluate an improved active error recovery scheme for reliable multicast (AERM). The AERM uses soft-state storage to facilitate fast error recovery. It has the following features: a simple NAK suppression and aggregation mechanism, an efficient hierarchical RTT measurement mechanism, an effective local recovery and scoped retransmission mechanism, and a periodical ACK mechanism. We implement the AERM and study its characteristics in NS2. We also compare performance with ARM and AER/NCA, both of which are representative active reliable multicast protocols. The results indicate that AERM can achieve considerable performance improvement with limited support from routers. Our work also confirms that active networks can benefit some applications and become a promising network computing platform in the future.

An efficient scheme for QoS information update among edge routers using multicast

It is essential and nontrivial to disseminate QoS information in an efficient manner and to guarantee the information accuracy. We present an efficient scheme for QoS information update among edge routers in a diffserv region. In our scheme, a multicast tree is constructed among the edge routers in a diffserv region to transmit the updates. This scheme is efficient in consuming network resources since the cost for QoS information update of using a tree is much less than using a flooding approach, and routers unrelated to decision-making are relieved from disseminating and storing QoS information. In addition, it is easy to implement our scheme in the Internet, since only edge routers need to deploy it. We investigate quantitatively the cost of our scheme for QoS information update against a flooding approach and study empirically the efficiency of several proposed QoS information update policies.