Network geography information systems,such as digital Earth,are getting more popular and offering geography navigation services for data requests of large-scale and high intensity.Hotmap is designed for Microsoft Virtual Earth by Microsoft Research to analyze the access distribution of spatial data.For the first time,Hotmap understands the request distribution of image data as power-law.However,Hotmap does not characterize the details.Through statistical analysis of four GlobeSIGht server logs with different granularity of times,the Zipf-like distribution(one of power-law)for image tile requests is found.Image data has a heavy skewed access,which means hotspot data exist and 20%(5%)of tiles pull nearly 80%(50%)requests.This finding helps the design and optimization of digital Earth system,for example,in aspects of image data storage declusting and distributed cache.

Network Geographic Information Service(NGIS)has been widely used in various types of projects,such as government decision making system,business navigation and disaster forecast.While NGIS is gaining in popularity,fast dissemination of voluminous and heterogeneous spatial data becomes a challenge due to limited Internet bandwidth and massive concurrent users.The spatial data should be delivered to the NGIS users within a reasonable time span.In this paper,a cache mechanism for terrain data and image data is proposed in NGIS to improve the interactive service performance.The tile cache index is designed based on the tile cache granularity,which is proved to be efficient.Tile Access average Interval time Longest(TAIL)cache replacement algorithm is proposed and its temporal locality and spatial locality are analyzed.The fast operation method of TAIL is presented.The image data set of LANDSAT7 with 30 meter resolution and the terrain data set of SRTM with 90 meter resolution are used in trace-driven simulations of TAIL.The result shows that TAIL has improved cache hit rate and byte hit rate,and significantly reduces tile-request-response-time and tile-request-number processed by NGIS server compared with currently available typical cache replacement algorithms.

Cooperative cache was introduced into the raster application server cluster to improve the QoS of raster data in GlobeSIGht. The trace-driven simulation based on server logs was performed. The optimal combination of number of application servers, cache size, cache replacement threshold and algorithm was figured out, which proved that cluster cache can largely save the I/O bandwidth from data storage. The test results of GlobeSIGht show that cluster cache can quickly response the tile requests.

Challenges such as limited simultaneous users and long tile request response times, exist in spatial information services. Cache mechanism is deployed in the services' client. The establishment, query and update of cache index for spatial data tile are analyzed. Tile cache lifetime excess and popularity replacement (TCLEPR) policy is proposed. In TCLEPR, cache tile has a longest lifetime excess, which is current cache life subtracted by average cache life, and the lowest popularity is replaced by new requested tile. Simulation results show that TCLEPR can improve tile request cache hit ratioes and byte hit ratioes, which in return shortens tile request response times and support more simultaneous users.

A distributed least cost algorithm is proposed by taking into account the waiting time of terrain tile request in server queue and the process time of it in server, based on an unit time cost vector. Dispatch probability space for each server is calculated and target server is determined through the landing point of a temporarily generated random number in the probability space. The scalability of the algorithm is satisfactory and the maintenance and update of clustered servers is flexible by configuring the unit time cost vector. Extensive simulations indicate that, for terrain tile requests of large-scale and high intensity, the least cost algorithm can balance servers' load efficiently and response the requests with minimal time. Simulation results also show that under low intensity requests, the considered load balancing algorithms behave approximately the same, while under high intensity request, it is better to design load balancing by considering queue situation of clustered servers.

An efficient and scalable call routing in multimedia system is particularly important in realizing optimum network resource utilization, supporting load balancing and improving QoS. Distributed and hierarchical routing architectures perform in large-scale multimedia networks not as well as in small and medium-sized packet switched networks. We study this problem and provide a scalable routing solution based on domain. From a routing point of view, large-scale multimedia networks comprise service domains and connecting domains, which are two basic domains for call process and call routing respectively. Next generation network is service driven, routing service is a special one provided to gatekeeper in service domain by next generation gatekeeper in connecting domain. Next generation gatekeeper is a call routing server, which extracts routing ability from gatekeeper and reinforces it to provide enhanced network-wide routing service flexibility with network investment savings. Routing information exchange criterions are defined for routing service creation and routing service provision. We describe how this application layer routing work better than others and confirm that its scalability and flexibility is suitable for practice in large-scale multimedia networks

The bandwidth problem is one of the most serious problems for wireless mesh networks due to the influence of interference. In this paper, we present a distributed interference-aware bandwidth guaranteed shortest path routing protocol in IEEE 802.11-based multichannel wireless mesh networks with admission control. We design an interference-free distributed TDMA schedule to guarantee enough link flow. The simulation results show our protocol achieves good performance that it can effectively provide bandwidth guaranteed path for connection request comparing with minimal hop-count routing protocol.

VoIP has been widely investigated to provide statistical quality of service (QoS) guarantees. We address the issue of how to select an efficient route to carry real-time traffic. Each eligible route is assigned a weight, which is updated according to network level QoS evaluation. When a new coming call arrives, originating signaling server first chooses two candidate routes from all practicable routes based on their weights. Then it executes endpoint measurement based admission control on the two routes to evaluate their QoS. The route with better performance is selected to carry VoIP traffic. Simultaneously, all weights of routes are self-adjusted based on the new collected network condition information. Simulation results confirm that weighted random selection based call routing algorithm achieves better statistical QoS guarantee than other methods in terms of end-to-end delay and call blocking ratio, with help of enhanced admission control.

Aiming at the problems that gatekeeper for video communication(called VCGK or Video Gatekeeper) can not support greater network scale and has the lower access efficiency in multi-zone, we study and analyze routing technology for largescale video softswitch from multi-zone and hierarchical system framework. We research on designing routing architecture reasonably, importing distributed agent technology and delaminating routing framework based on policies. By using TRIP protocol for communication between agents, we deal with the update and broadcasting of call routing information concentratedly, converge routing info, predigest routing process, and solve the key problems of call control and connection control about real-time video communication. Considering routing QoS and network load balance, we put forward a dynamic routing transfer policy based on distributed routing agent. Finally this paper forms an economical and effective call routing addressing solution which can meet the requests of next generation largescale multimedia video communication network, and solve the difficulty of location in H.323 system appropriately. It can help to enlarge the scale of video communication network based on H.323 protocol.