No abstract

per Erlang (Watts/Erlang) for a voice system, or Watts per bit per second (Watts/[bit/s]) for a data system. Other candidates that incorporate area covered include Watts/Erlangs/km 2 for voice or Watts/ (bits/sec)/km 2 for data systems. While these metrics may appear to be a natural way of assessing how effectively power is used [5], we believe that they are not appropriate for a wireless system. In fact, they may be counter to prudent system operation as well as counter to a "green" goal. In addition, focus on these metrics could lead to the questionable conclusion that reduction of total radiated power in current networks constitutes a favorable tradeoff of "greenness" for performance.

There is considerable interest within the industry in specifying the performance of wireless data. In this letter, we review and contrast various performance metrics such as peak rates, average throughput, and user experience. We discuss the relevance of these measures to varying applications such as marketing, planning, and technology comparisons.

Insights into this tradeoff are crucial for facilitating optimal expenditure of service provider resources. This point is illustrated by the dilemma of deploying additional carriers within an Evolution-Data Optimized Revision A (EV-DO Rev. A) system. EV-DO Rev. A is a time-shared wireless data system where the full power and full bandwidth of a forward link wideband carrier is devoted to one user at a time. A scheduler decides which user to serve within each time slot. The decision is made based on multiple sources of information, including the radio channel