Internet traffic engineering is much important for Internet Service Providers (ISPs) today, since it can be used to fully utilize already deployed network resources. For ISPs, the requirements for traffic engineering should be simple, easy to configure, cost-effective and efficient. Based on these considerations, we propose an algorithm called Web First Adaptive Traffic Engineering (WFATE). Since World Wide Web (WWW) services dominate most of the total Internet traffic and WWW flows are not long-lived, we only apply load balancing to WWW traffic in the algorithm. It can be shown that the number of coexistent WWW flows at an ingress node is almost certainly below a bound, and thus a forward-per-flow mechanism without keeping track of the state of each flow is feasible. This mechanism can balance traffic load at fine granularity and therefore get better performance. Through simulations and performance comparison, it is shown that WFATE is quite efficient, which can improve the network throughput averagely by 26% under the "dense source" traffic pattern and 9% under the "sparse source" traffic pattern.

The use of cross-polarized antennas for multiple-input multiple-output (MIMO) systems is receiving attention as they are able to double the number of antenna for half antenna spacing needs. This paper presents the channel correlation property of the 3rd Generation Partner Project (3GPP)/3GPP2 spatial channel model (SCM) with the polarization propagation. The statistical average of the per path polarization correlation given random cross-polarization discrimination (XPD) with co-located ideal tilted dipole antennas is derived. The impact on the random behavior of the polarization correlation due to the slant offset angle, the per path angular spread (AS), and the random XPD is analyzed. The simulation results show that the variation of polarization correlation caused by the random XPD is maximized with a 58 slant offset angle under the assumptions of all predefined scenarios in SCM. The per path AS has minor impact on the statistics of the polarization correlations. The randomness of polarization correlation is negligible for an XPD with small standard deviation.

A number of on-line and off-line algorithms for load balancing on multiple paths for MPLS (MultiProtocol Label Switching) traffic engineering have now been proposed, in which it is always assumed that sets of LSPs (Label Switched Path) have already been established between node pairs. While how to choose these paths is an important issue in traffic engineering, it has not been well studied yet. In this paper, we attempt to fill in this gap. As the shortest paths are always preferred in routing problems, we evaluate several k shortest path algorithms from the viewpoint of bandwidth use efficiency and the number of the found paths. Extensive simulations have been performed in different kinds of topologies to factor out effects of network characteristics on these algorithms' path calculation performances. It is found out that the performances of the evaluated algorithms are limited in some cases and the design of new algorithms for the path calculation problem is worth studying in the future.

Multipath is a big problem for TCP traffic in traffic engineering. To solve it, hash functions such as CRC-16 are usually applied over source and destination address segments in packet headers. Through simulations and performance comparison of several multipath algorithms, it is found out that high network utilization achieved by using hash functions is at the expense of low fairness among coexisting TCP flows. It is also illustrated that packet size has significant influence on performance.