Traditionally, cellular systems have been built in a hierarchical manner: many specialized cellular access network elements that collectively form a hierarchical cellular system. When 2G and later 3G systems were designed there was a good reason to make system hierarchical: from a cost-perspective it was better to concentrate traffic and to share the cost of processing equipment over a large set of users while keeping the base stations relatively cheap. However, we believe the economic reasons for designing cellular systems in a hierarchical manner have disappeared: in fact, hierarchical architectures hinder future efficient deployments. In this paper, we argue for completely flat cellular wireless systems, which need just one type of specialized network element to provide radio access network (RAN) functionality, supplemented by standard IP-based network elements to form a cellular network. While the reason for building a cellular system in a hierarchical fashion has disappeared, there are other good reasons to make the system architecture flat: (1) as wireless transmission techniques evolve into hybrid ARQ systems, there is less need for a hierarchical cellular system to support spatial diversity; (2) we foresee that future cellular networks are part of the Internet, while hierarchical systems typically use interfaces between network elements that are specific to cellular standards or proprietary. At best such systems use IP as a transport medium, not as a core component; (3) a flat cellular system can be self scaling while a hierarchical system has inherent scaling issues; (4) moving all access technologies to the edge of the network enables ease of converging access technologies into a common packet core; and (5) using an IP common core makes the cellular network part of the Internet.

In this work, we evaluate the performance of voice over Internet protocol (VOIP) services over high speed downlink shared channel (i.e. UMTS-HSDPA), that is used in third generation cellular systems to provide a high speed packet data service. We focus on examining system outage based on delay budget in radio access networks. We conclude that VOIP system performance is highly sensitive to delay budget. In order to support VOIP services with end-to-end delay comparable to that of circuit switched voice, the HS-DPA system needs to be carefully engineered, and design alternatives that relax the delay budget in radio access networks could be considered.

The wireless industry is undergoing a transition. We see the evolution from cellular 2G to 3G standards, the migration from circuit to packet applications, and the procession of voice to data. We also see the industry incorporating new wireless access technologies such as WiFi and WiMAX. All of this is occurring in a market place where voice subscriber penetration levels in many parts of the world are saturating and there is incredible pressure to reduce network capital and operating costs. These conditions present significant technology challenges to increase customer demand and further increase network capacity to lower network costs. In this talk we will review the market drivers and applications that will drive the next-generation services. We note that in response to these the network architecture will migrate to having more control at the edge with substantially more direct base station to base station communication. We will then present several technologies that address these issues.

We study in this paper whether mobility can be supported by the 802.16d network without any change in the specification. Mobility capability involves two main issues: connection handoff and correct reception for moving terminals. We find that seamless connection handoff can be achieved within the 802.16d standard by: 1) applying some of the existing functionalities defined for the terminal initialization process; 2) devising a set of protocols for message exchanges for handoff; and 3) forwarding some of the operational parameters from the current base station to a new one via the backhaul network, instead of over the radio link. As for reception at moving terminals, our analysis of bit error rate for the OFDMA mode shows that, under typical radio conditions, the link can provide satisfactory error performance for terminal speed up to tens of kph. As a result, we show that the current 802.16d standard with our proposed technique can support user mobility.

Third generation code-division multiple access (CDMA) systems propose to provide packet data service through a high speed shared channel with intelligent and fast scheduling at the base-stations. In the current approach base-stations schedule independently of other base-stations. We consider scheduling schemes in which scheduling decisions are made jointly for a cluster of cells thereby enhancing performance through interference avoidance and dynamic load balancing. We consider algorithms that assume complete knowledge of the channel quality information from each of the base-stations to the terminals at the centralized scheduler as well as a two-tier scheduling strategy that assumes only the knowledge of the long term channel conditions at the centralized scheduler. We demonstrate that in the case of asymmetric traffic distribution, where load imbalance is most pronounced, significant throughput gains can be obtained while the gains in the symmetric case are modest. Since the load balancing is achieved through centralized scheduling, our scheme can adapt to time-varying traffic patterns dynamically.