Should operators deploy a macro or micro cellular architecture in 2.5 GHz band? This is not a trivial question particularly for an incumbent wireless operator that already holds spectrum in lower spectrum bands. The same can be said of Greenfields looking to capitalize on the exponential demand for wireless data services.
The main issues concerning wireless networks in higher spectrum bands such as 2.5 GHz is propagation characteristics. This is an important point, because from a system perspective, higher frequency systems have greater system gain. Yet the advantage in system gain is erased when poorer propagation characteristics are factored in. For example, a 2.6 GHz system would typically have about 11 dB higher system gain than an 800 MHz system. But if we consider wall losses for in-building service and the added attenuation because of the higher frequency, the result is at least a 5 dB loss in the maximum allowable path loss which translates into 25% reduction in cell radius over a comparable system in 800 MHz and consequently in 44% reduction in cell size. In short, it would take about double the number of cells to build a network in 2.6 GHz than in 800 MHz. Factoring capex and opex over the lifetime of a network, it’s no surprise that the price of 2.6 GHz spectrum is lower by 5-35x than that of 800 MHz spectrum.
|Gain Difference (Loss is negative)||2600 over 800 MHz|
|eNodeB Antenna||4 dB|
|eNodeB Cable||-1 dB|
|UE Antenna||5 dB|
|UE Sensitivity||3 dB|
|System Gain – Difference||11 dB|
|Wall Penetration||-6 dB|
|Frequency Dependence (20logf)||-10 dB|
|MAPL Difference||-5 dB|
|Distance (2600 < 800 MHz) (35logD)||25%|
|Size (2600 < 800 MHz)||44%|
Statistics on wireless usage models show that at least 40% of traffic comes from in-buildings. This means that the density of cells in 2.6 GHz has to be high to reach similar performance inside buildings to that of 800 MHz, for example. Since this increases interference and requires a high frequency reuse factor, I don’ think operators would favor a macro-cell deployment model unless they own a significant chunk of spectrum (as does Clearwire, for example). Hence, I tend to favor deployment of small cells in this band to complement macro-cells operating in lower frequency bands. This has the advantage of eliminating the need for macro/micro-cell coordination.