Distributed dynamic resource allocation for multicell SDMA packet access net

Abstract

This paper addresses the issues of distributed dynamic resource allocation (DRA) in the downlink of a SDMA broadband wireless packet network with multiple access points and adaptive antennas. Packet access and downlink beamforming make the intercell interference hard to predict, and how to handle it in order to efficiently perform slot and power allocation is still an open issue. We propose here new DRA strategies for efficient allocation in a distributed environment, which jointly consider spatial compatibility of users and an intercell interference estimation based on a fraction of the worst case interference measurements. The latter is independent of the actual allocation. We consider a low mobility TDD system with synchronized base stations and we assume perfect channel knowledge. The algorithms which are introduced and compared are the distributed max-min fit (DMMF) and the distributed reverse fit (DRF), which are based on the max-min fit criterion, the DMMF combined with the novel concept of nulls preallocation, where each AP reserves some beamforming nulls for the most interfered users of neighboring cells, and the power shaping technique (PS-DRA for SDMA), which efficiently exploits static power preallocation on time slots. The numerical results show that our techniques are able to significantly reduce the gap between a greedy centralized strategy that fully coordinates all the access ports and the baseline case of random distributed allocation, previously proposed for packet access. However, distributed power control in a packet switched environment with downlink beamforming still remains a hard task: nevertheless, PS-DRA is an efficient solution for joint slot-power allocatio

Dynamic Time-Frequency Division Duplex

“…Distributed D-TDD schemes for BSs have been investigated in [18], [19], [21]- [25] and references therein. In particular, [18] proposed cooperation among cells that resembles a centralized D-TDD scheme.…”

“…The authors in [19] proposed a D-TDD scheme which alleviates the ICI impairment by splitting the uplink and downlink frequency. Authors in [21] and [22] investigate a distributed D-TDD scheme designed for multiple-antennas. The work in [23] proposes a distributed multi-user D-TDD scheduling scheme, where the ICI is not taken into account, which may lead to poor performance g replacements in practice.…”

“…We note that [13], [18], [19], [21]- [25] require full knowledge of the ICI, which may not be practical, as discussed in Sec. II-D. We also note that the schemes in [13], [18], [19], [21]- [25] transform to the static-TDD and/or static-FDD scheme when ICI is not known. Hence, the static-TDD and/or static-FDD are much more practical for implementation than the D-FDD duplexing scheme since they do not require ICI knowledge.…”

“…We note that [13], [18], [19], [21]- [25] require full knowledge of the ICI, which may not be practical, as discussed in Sec. II-D. We also note that the schemes in [13], [18], [19], [21]- [25] transform to the static-TDD and/or static-FDD scheme when ICI is not known.…”

Optimal Centralized Dynamic-Time-Division-Duplex

“…Now, note that the only variables that can be manipulated in (19) in each time slot are the values of the elements in the vectors t(i), r(i), and f(i), and the values of µ k , ∀k. We use t(i), r(i),…”